Election Day is nearly here, and with dozens of competitive races for House, Senate, and governor on the ballot, it’s going to be a doozy. Using the map below, you can explore when polls close in the most competitive House and Senate races. Note that results usually start coming in 5-15 minutes after polls close.

If the competitive races are too much to follow, click the “likely tipping point” button at the top of the map to highlight only races most likely to decide control of Congress. For more detailed maps, explore our county-level benchmarks here, which will be updated with actual results when they come in. For written analyses on election night, check out the live blog in our notes section:

Notes (live blog will be here)

But how should we interpret the results as they come in, and when might we expect key races to be called? Let’s break it down hour-by-hour.

🕕 6 pm ET

Polls close in most of Indiana and Kentucky

Read more

Our live results blog will be here

Bonus features

Subscribe for free (no existing account needed) to get the map as a downloadable table. You’ll also get access to our forecast, which we’ll run throughout the night, and future articles and models for the 2027 and 2028 elections.

Read more

Earlier this week, Emerson College surveyed voters in ten congressional districts, all of which voted for President Trump in 2024. None are particularly likely to decide control of the House of Representatives. Instead, they represent the upside for Democrats if they are to win a comfortable working majority, similar to what they had after the blue waves of 2018 or 2008. The findings were unequivocally bullish on Democrats, who led in nine of the ten polled districts.

Emerson isn’t the only pollster releasing bad results for the House GOP. Let’s take a look at the 16 Trump-won districts polled by independent firms in the last month. If their surveys are correct, these districts have shifted an average of 15 points towards Democrats since 2024.

Read more

This week, candidates across America filed their third-quarter fundraising reports, covering July 1 to September 30—the last full quarter before Election Day. Raising money has historically correlated with electoral strength, so the Q3 numbers are a key indicator of momentum heading into the midterms. Let’s take a closer look at which candidates did best.

A big quarter for Democrats (outside of Minnesota)

In our Senate forecast, 18 states are less than 99% certain for either party. Democratic nominees raised more money in all but one of those contests, racking up a whopping $254 million to Republicans’ $55 million.1

Topping the list is Democratic nominee for Texas Senate James Talarico, whose $60.2 million haul is the largest in US history. Another noteworthy mention is Angie Nixon, the Democratic nominee in Florida. She raised less than one million dollars before July, but pulled in $7.2 million in Q3, far exceeding her opponent.

What makes these figures more surprising is the presence of nine Republican incumbents on the list. Incumbents tend to enjoy a financial edge, yet every GOP Senator here is being outraised. Kansas’s Roger Marshall is in particularly dire shape, raising 12 times less money than his challenger. That’s one reason why Kansas, a state Donald Trump won by 16% in 2024, is competitive.

The GOP’s one bright spot is Minnesota, where Republican nominee Michele Tafoya raised $4.3 million to her opponent’s $3.8 million. This is one of the smaller gaps on the list, but it’s the only one favoring a Republican, and polling has shown a surprisingly close race in the North Star State.

17 more Senate races are totally uncompetitive in our forecast, but the main party nominees still pulled in more than $23 million across them. As expected, there isn’t much drama on this list. The most eye-popping figures come from Arkansas and Idaho, where a few polls have shown close races. Illinois stands out too, but only because Republican nominee Don Tracy loaned himself $1.5 million.

How this compares to past cycles

Senate Republicans are used to being outraised—the last time they pulled in more money than Democrats was 2016. Instead, it may be helpful to look at how each party is doing relative to the last competitive Senate race in each state.

Democrats are doing better this year in Florida, Alaska, Texas, and especially South Carolina. In 2020, both candidates in the Palmetto State’s Senate race raised over $100 million, but in Q3, appointed Republican Senator Darline Graham2 pulled in less than $2 million. Her opponent raised $9.4 million.

Republicans’ biggest improvements are in Nebraska, where they learned to take independent Dan Osborn seriously after 2024, and Maine, where Democrats had to replace their nominee at the last minute. Republicans are also in better shape overall in Iowa and Kansas, but because Democrats had a late fundraising surge, both are in play.

Still, raising money isn’t everything. Democrats in particular have done so and lost in the past, notably in Kansas in 2020 and Ohio in 2024. Both were fairly red states, and the GOP offset their fundraising deficits with outside spending. But both factors are relevant to this cycle as well: the most competitive Senate races are once again in red states, and Republicans have the edge in ad spending.

Even so, as the national environment shifts away from the GOP, grassroots fundraising could decide the closest races. Based on 538 research, a candidate who receives 60% of contributions in a race should do a point better than one who receives 50% (all else held equal). Our Senate forecast currently shows three races—Alaska, Texas, and Iowa—being decided by about that much.

With three weeks to go until Election Day, Democrats could win a majority of governorships for the first time since 2010. Because gubernatorial races are less polarized than congressional ones, the playing field is broad, encompassing everything from deep blue Oregon to solid red Nebraska. Without further ado, let’s dive into the latest trends in our polling averages.

Share

Texas is the biggest competitive governorship up this cycle, and polls show an upset brewing between three-term Republican incumbent Greg Abbott and Gina Hinojosa, a Democratic state legislator. Initially seen as a shoo-in for reelection, Abbott entered the cycle with a 20-fold cash advantage, but now leads the polls by just 2%. Should the broader trend since July—undecided voters breaking for Hinojosa—continue, Republicans may be in for a Texas-sized shock.

Florida, where Republican governor Ron DeSantis is retiring after two terms, is looking less shaky for the GOP. Democrats are running former Republican Congressman David Jolly, while the GOP nominated current Congressman Byron Donalds, who enjoys a consistent, if sometimes narrow, lead in the polls.

Democrats are ahead in the polls to flip one big governorship: Ohio, where incumbent Republican Mike DeWine is term-limited. In the race to replace him, Democratic nominee Amy Acton has a 3-point edge over Republican billionaire Vivek Ramaswamy, capturing 48% of the vote to Ramaswamy’s 45.

Meanwhile, in neighboring Iowa, Democrats may have their best pickup opportunity of the cycle thanks to state auditor Rob Sand, the last Democrat to win statewide. He leads in our average by more than 7 points, which could drag Democrats over the finish line in 3 of the state’s 4 House seats.

In a friendlier environment, Republicans might’ve been focused on making gains in the “Blue Wall” states that decided the last presidential election: Wisconsin, Michigan, and Pennsylvania. The closest of the three is Wisconsin, where Democrats nominated Milwaukee executive David Crowley in a tight primary. His opponent, GOP Congressman Tom Tiffany, is down 4% in the polls. If that holds, Democrats will be poised to win their first trifecta in the Badger State since 2010.

In Michigan, Republicans were hoping to gain a seat after the retirement of governor Gretchen Whitmer. Their nominee, Congressman John James, ran strong Senate campaigns in 2020 and 2018, but is struggling against Democratic secretary of state Jocelyn Benson, who has a track record of winning statewide.

Lastly, Pennsylvania is taken off the board by popular Democratic governor Josh Shapiro, who leads our polling average for the Keystone State by a whopping 21 points. Democrats hope his coattails will be enough to flip the state’s Senate, delivering their first trifecta in more than 30 years.

One swing state that is not breaking in Democrats’ favor is Georgia. After leading in early polls, former Atlanta Mayor Keisha Lance Bottoms has fallen a point behind Republican billionaire Rick Jackson. Jackson is self-financing his campaign and holds a massive spending advantage.

Republicans also have the financial edge in Nevada, where Republican governor Joe Lombardo faces the state’s Democratic attorney general, Aaron Ford. While Lombardo has led nearly every poll conducted so far, a September survey from Emerson College found Ford up 2%, suggesting a competitive race.

In neighboring Arizona, Democratic governor Katie Hobbs holds a relatively comfortable advantage after winning in a nail-biter four years ago. She’s led in every survey since early 2025, and is up nearly 8 points on average.

Further up the West Coast, Republicans look surprisingly close to picking up the governorship in Oregon, where Democratic incumbent Tina Kotek is widely unpopular. Even so, she just took the lead in our polling average, suggesting Oregon’s blue tilt could rescue her once more.

That leaves Republicans with one straightforward pickup opportunity: Kansas. Democratic governor Laura Kelly won by narrow margins in 2018 and 2022, and her retirement leaves the seat wide open. Republican Ty Masterson leads our polling average by just under 4%.

Democrats, meanwhile, got a brand new flip opportunity this summer in Alaska, which was initially seen as a likely GOP seat. That’s because three Republicans advanced to the general election with only one Democrat, splitting the vote. While ranked-choice voting (RCV) does let voters rank multiple candidates, polls show that RCV may not be enough for the Alaska GOP.

Finally, we’ll be watching for upsets in Nebraska and Vermont, both states that weren’t on the map at the start of the cycle. In Nebraska, Republican incumbent Jim Pillen has to contend with Democrat Lynne Walz and Brett Lindstrom, a Republican legislator running under the America First Party. Our polling average shows Pillen ahead of Walz by just 3 points.

Meanwhile, in Vermont, the bluest state in the country, Republican governor Phil Scott has historically cruised to victory on his moderate brand. Recent polling has shown the state reverting towards its blue roots, however, and Scott only has a 3% lead in our aggregate, a far cry from his 52-point win two years ago.

In summary, if the polls are right, Democrats are on track to flip Alaska, Ohio, and Iowa, while Republicans would take Kansas. That nets Democrats two seats, giving them 26 of America’s 50 governorships. If they beat the polls by a few points, they could also pick up seats in Texas, Georgia, Nevada, Nebraska, and Vermont. If Republicans beat the polls by a few points, however, they’d be favored to hold 25 governorships.

This article is part of a series analyzing the donations and political dynamics surrounding the most important Senate races of the 2026 midterms. It is done in collaboration with Frontline Democracy. You can read our overview of Alaska’s Senate race here, and analyses of Nebraska and Ohio over at Frontline Democracy.

When Donald Trump defeated Kamala Harris in 2024, he won Alaska by more than 13 points. That wasn’t exactly a surprise. In fact, only one Democratic presidential candidate has won Alaska since it became the 49th state: Lyndon B. Johnson in 1964.

Alaska has a ruby-red history, and after Trump’s 13-point victory, Republicans seemed confident about holding the seat in the 2026 midterms. Two-term incumbent Dan Sullivan’s (R) decision to run for re-election reinforced the widespread belief that Alaska would be off the table for Democrats, even in a good year. Given the state’s partisan lean and Sullivan’s unproblematic profile, the Cook Political Report initially rated the race as “Solid Republican.”

A month ahead of the midterms, all that seems to have changed. Not only is Alaska competitive; according to our forecast, Democrats are now narrowly favored to win the seat, which could put them ahead in the battle for the Senate. The monumental shift has one clear cause: Democratic nominee Mary Peltola.

Share

Peltola Leads with Raw Donors

Peltola, an Alaska native, entered the political stage by flipping the Last Frontier’s statewide House seat in a 2022 special election and holding it in a difficult midterm year for Democrats. In 2024, she was one of only 16 House Democrats who outran Kamala Harris by over 10 points. Her popularity has translated into in-state fundraising strength for her Senate campaign.

Peltola swamps Sullivan among in-state individual donors by 40 points. In total, she has received contributions from 2034 unique donors, compared to 863 donors for Sullivan. I should note that this data includes itemized donations, meaning only contributions by individuals who donated over $200 across all races in this election cycle were counted.

The Democrat’s commanding lead stretches across the entire state but is strongest in ZIP areas in Anchorage and Juneau. The few scattered pockets where Sullivan leads are relatively small, as those only have one or two total donors. Almost all of Sullivan’s donors are located in Peltola-led ZIP codes.

But raw donor numbers don’t decide the money race.

A Close Cash Race in Alaska, but Republicans See Strong Out-of-State Spending

The problem for Peltola is one Democrats know all too well. While Democrats often excel at building a network of individual donors, it’s Republicans who receive the bigger checks. Peltola is beating Sullivan with raw donors, but the incumbent’s average donation is almost 2x bigger than Peltola’s, creating a more balanced fundraising map. You can explore our national fundraising and donor maps with data for 15 key 2026 Senate races here:

Frontline Democracy

Interactive Donor Maps: Explore the 2026 Senate Fundraising Performance

Read more

2 months ago · 2 likes · Clemens Stein

Driven by these higher contributions, quite a few ZIP codes in and around Anchorage now flip to Sullivan, breaking the Democrat’s dominance in Alaska’s biggest population hub.

Still, Peltola’s lead among individual donors proved strong enough to fend off Sullivan’s larger donation, as she narrowly leads him with raised in-state funds. The influence of these in-state fundraising tallies may, however, not matter much in the end.​

More Donors, More Votes? A look at Alaska’s Primary Results

In Alaska and every other battleground race, out-of-state money does most of the heavy lifting, which is why Republicans currently have a clear edge in ad buys for Alaska’s Senate race according to AdImpact. As of September 11th, the GOP had reserved $23.1M in future ads for Sullivan, dwarfing Democrats’ planned $7M worth of ad buys.

While money will undoubtedly decide who dominates the voters’ airwaves in the next month, a lead as significant as Peltola’s individual donors can indicate excitement on the ground, which will be just as decisive come November. Thanks to Alaska’s jungle primary system, we already got a glimpse at what that excitement could look like, and how much of a predictor our donor map could be.

Peltola finished first with 49.5% in the August primary, eight points ahead of incumbent Dan Sullivan. Sullivan’s vote share was diluted by other Republicans, but even the two-party vote share painted a poor picture for the incumbent. After you combine all primary votes for Republican candidates and stack them up against all the votes the Democrats received, Democrats end up ahead.

Our analysis shows that, in the primary, Peltola performed best in areas where she holds a larger individual donor share lead, pointing to a correlation, albeit a weak one (R²=0.204). She won 12 out of the 13 state house districts where she had at least 40% more donors.

We used Peltola’s lead margin among donors as the basis for this analysis because she leads among donors in almost every district across Alaska. Assessing the predictiveness of our donor map still proved difficult. Primary results are recorded by state house district in Alaska, while donor data is recorded by ZIP code. We tried to pair every ZIP code with a state house district to the best of our abilities, though some ZIP codes map to multiple districts.

Excitement and money aren’t everything, but both can play a crucial role in tilting close races. As of today, the race for Alaska Senate is a dead heat, with Mary Peltola at 49% and Dan Sullivan at 46.1%. That number almost exactly mirrors her primary result. Alaska has a ranked-choice voting system, meaning a candidate needs to clear 50% to win. Peltola is close to that threshold in the polls.

On November 3rd, two factors may propel her to victory. First, Alaska’s general elections were more democratic than primaries in past midterms, a trend that might repeat itself. Second, ballot exhaustion among the other two marginal candidates – both Republicans, including a second Dan Sullivan – may put her over the top. It’s still an incredibly close race, but as of today, you’d rather be in Peltola’s shoes than in Sullivan’s.

Who’s winning the midterms?

Every news outlet in America will ask some variation of this question in the next five weeks. Some will answer it with polling data—candidate X leads candidate Y by this many points—but it’s important to remember that polls can get it wrong. In 2024, the average poll was off by 3.3%, and that was considered a good year for the polling industry.

Share

How much error should we expect?

Polling error can depend on which state you’re looking at. Based on Silver Bulletin’s collection of polls taken in the last three weeks of every election cycle since 1998, the average historic error has ranged from 3-4% in states like Georgia and North Carolina to over 10% in Wyoming and Hawaii.

Some of this could come down to random variance or local effects—which pollsters are active in a state, candidates with unique coalitions, last-minute scandals—but states with more polling error seem to have two things in common: unique demographics1 and not being polled often.

Combining the aforementioned variables (polling intensity and demographics), I built a simple model of average error in every state. The results should help set expectations for the states with competitive races this fall:

As you can see, most competitive states sit at the lower end of the error range: about 4% on the margins. In Georgia, for example, if you see a candidate up by 4 points, a normal polling error could turn that into a tie or an 8% blowout.

Related: Which Party Is Winning the Senate in 2026?

Not every competitive state will fall into this range of error, however. Maine, home to one of the nation’s most competitive Senate races, has the highest modeled error and second-highest observed error of the competitive states. A six-point polling miss would be historically average there.

Iowa and Alaska, both home to competitive gubernatorial and Senate races this fall, have average polling misses of nearly 6%. Kansas, which is competitive at the Senate level, tops the list with an average error of over 7%.

Now the polls aren’t guaranteed to be inaccurate in those states this cycle, as the polling industry can have good and bad years. And because it’s hard to predict who will benefit from polling error in advance, you’d still rather be the candidate in the lead. Here’s another way of thinking about it: the more accurate polling is, the more likely the leading candidate is to win.

With less than 50 days to go until Election Day, Democrats are within striking distance of flipping the Senate. The Republican majority is formidable on paper, requiring Democrats to flip multiple red states to retake the chamber. Since Labor Day, however, polls have shown just that: Republicans trailing in states President Trump carried by 11, 14, and even 20 points in the last election.

In Iowa, a Trump +13 state, polls show Democratic state legislator Josh Turek neck-and-neck with GOP Congresswoman Ashley Hinson, a strong fundraiser with a history of outrunning Trump in her district. Recent polls have ranged from Turek +5 to Hinson +4, pushing our polling average to an effective tie.

In Texas, a state President Trump carried by nearly 14% in 2024, state Attorney General Ken Paxton snagged the Republican nomination in a tough primary this spring. Paxton has trailed in 16 of the last 20 general election polls, and his opponent, Democratic state legislator James Talarico, is outraising him by a factor of 8. By comparison, Democrats had a 2-to-1 fundraising edge in Texas’s last competitive Senate race in 2018, and trailed in most polls.

While Texas and Iowa are open seats, multiple sitting GOP Senators are also at risk. In Alaska, Republican Senator Dan Sullivan trails former Democratic Congresswoman Mary Peltola by 1.5%. In Ohio, former Democratic Senator Sherrod Brown leads incumbent Republican Jon Husted 48 to 45. And in Nebraska, a Trump +20 state, incumbent Republican Pete Ricketts trails independent Dan Osborn 46 to 47.

Related: 2026 Senate poll tracker

Looking to the traditional swing states, Republicans seem poised to lose North Carolina, the perennially close state where Democrat Roy Cooper now leads by 8 points. Maine, one of the few competitive blue states this year, is kept close by Republican Senator Susan Collins, arguably the toughest incumbent in the country. She trails former Maine Senate President Troy Jackson by 2 points. And while the GOP might have had several targets of their own in a better year—Michigan, Georgia, Minnesota, and New Hampshire—polling shows Democrats successfully defending all four so far.

The case for Republicans

If the midterm battleground truly cuts this deep into Republican territory, a blue Senate isn’t just plausible, but probable. Taking the polls at face value, we’d expect the GOP to win 46 Senate seats this fall, down from the 53 they hold today—the worst loss either party has suffered since 2014. But could Democrats’ advantage evaporate by Election Day?

Nationwide polling does suggest a strong midterm environment for Democrats, but not as strong as state polls. Based on each state’s partisanship and the generic ballot—national polls asking voters to choose between generic candidates from each party—we’d expect Democrats to win 49 seats, two short of a majority. This is consistent with Split Ticket’s state-level generic ballot and NYT polling.

In other words, Democrats (and an independent) are currently ahead in 5 seats a generic candidate from their party “shouldn’t” win. Systematically beating the generic ballot isn’t impossible—Democrats have enjoyed a candidate quality advantage in recent elections, especially the 2022 midterms.

Related: How Accurate Are Early Senate Polls?

Candidate quality can be exaggerated by early polling, however. We saw this in 2024 , when popular GOP governor Larry Hogan led in early polls of the Senate race in Maryland, a blue state. Democratic Senators in the red states of Ohio and Montana also led early that cycle. All three lost after voters came home to their respective parties.

The middle of September isn’t exactly early, however. Based on the time left and number of undecided voters in each state, our forecast model thinks that Republicans are more likely than not to catch up in Iowa and Nebraska, where the polls are close, and the electorates historically lean Republican. That gets the GOP to 48 seats. Alaska, Ohio, and Texas, the other red states where Democrats lead, are toss-ups. If partisanship overwhelms Democrats in two of those states, or if Republicans win in Maine and Michigan (where our model sees them as competitive underdogs), the Senate stays red.

The catch: things won’t necessarily get better for the GOP

Democrats have their own reason for optimism: the president’s party tends to fall further behind as Election Day closes. We’ve collected polls covering every midterm election since 1994, and the president’s party only gained ground in one cycle. In 2002, when President Bush’s approval rating was above 60%, Republicans gained about 2 points nationally between September and November. In 2006, however, when Bush’s approval was in the 30s, Republicans fell about 3 points.

Today, President Trump’s approval rating is closer to Bush 2006 than 2002, and Democrats have been gaining in the polls all cycle. Gas pipeline disruptions from the Iran War provide one plausible path to further deterioration. Pollsters switching to polling likely voters, who are expected to lean more Democratic this year, provide another.

The effects of a state-level Republican rally in red states, combined with a national bump for Democrats, are a wash: our overall forecast has the Senate as an effective toss-up.

A close forecast does not guarantee a close Senate in November, however. It’s possible that the polls will underestimate one party in almost every state, as happened in 2020, and both parties could make a clean sweep of the swing states. Accounting for this, our forecast gives a nearly 1-in-3 chance of more extreme scenarios happening: Democrats not netting any seats, or netting more than six.

On Thursday evening, the Supreme Court of Missouri unanimously blocked the state’s 2025 congressional map. As part of a broader effort to help Republicans hold the House in the midterms, the new map would’ve made the Democratic-held 5th district an easy Republican pickup and shored up the vulnerable GOP-held 2nd district. Instead, barring an intervention by the US Supreme Court (SCOTUS), the map will be put to a statewide vote this November. In the meantime, any House elections will be held under the less Republican-favorable 2022 map.

Share

How this changes the battle for the House

Republicans are entering the midterms with 220 House seats, the narrowest majority in nearly a century. With such a thin buffer—they can only afford to lose two seats—holding the chamber this November will be an uphill battle. Midterm electorates tend to swing against the party in power, and 2026 looks no different. As such, even a great night for the GOP will likely see the House come down to a few seats, and the Missouri ruling costs them dearly in two.

Using the 2024 election as a benchmark, returning to the old 2022 map makes the light red 2nd district 4% bluer. The Kansas City-based 5th district gets a whopping 42% bluer. Our midterms model, which forecasts the election using the 2025 map, currently expects Republicans to win seven of Missouri’s eight House seats. Under the old map, they’d win just five or six. The most obvious beneficiary of the change is MO-05’s Democratic incumbent, Emanuel Cleaver, whose odds of winning go from 15% to 100%.

Perhaps more interestingly, the 2nd district would immediately become one of the most competitive races in the country—one of roughly 15 seats that get the “toss-up” rating in our forecast. Democrats’ odds in the district more than double from 18% to 38% under the old boundaries.

Per our calculations, incumbent Republican Ann Wagner put up subpar performances in both the 2024 and 2022 elections. Split Ticket corroborates this, estimating that she underperformed in 4 of her last 6 races. Wagner did win comfortably in 2024 and 2022, but against poorly funded opponents in R+2 national environments. The national environment is closer to D+7 this year, so the district’s R+6 partisan lean would leave a “generic” Republican underwater.

Our model does consider the boost incumbents generally receive, however, which pushes Wagner up about 3%. Factoring in candidate quality pushes her down a point. In the end, we’d expect the seat to remain Republican, but a Democratic upset would be well within the realm of possibility.

Could Republicans use the new map anyway?

Missouri Republicans aren’t going down without a fight. The state attorney general responded quickly to the map’s blockage, declaring her intent to appeal to SCOTUS. The ruling won’t be easy to overturn, however, as justices unanimously held that Missouri’s Constitution allowed a referendum on the map. That conclusion could be supported by SCOTUS precedent, particularly Ohio ex rel. Davis v. Hildebrant (1916), which upheld another referendum on a congressional map.

However, primary elections were already held on August 4th using the new congressional boundaries, creating an unprecedented situation. Implementing the old map means candidates nominated under one set of districts will be elected by another. Republicans could argue this angle, asking SCOTUS for a temporary stay to keep the map in place for November without declaring the referendum illegal. Missouri Democrats are certainly in a much better position than they were two days ago, but the fight is not over.

NEW: Senate and House forecast

Other averages

Governor / Senate / Data centers / Presidential approval / Generic ballot

Methodology

US politics are all about public opinion, and there’s no better way to capture that than a well-designed poll. Unfortunately, as we’ve covered extensively here at PlusMinus4, polls are prone to error. This typically results from not sampling enough people, or methodological shortfalls that result in a non-representative sample.1 Averaging together a large number of polls minimizes both causes of polling error, though it doesn’t eliminate either entirely.

1. Using polls

We include polls as long as they include the basic information our models need: who conducted it, when it was conducted, and which population was polled. We do have some rules for when pollsters release multiple versions of their surveys:

• Population: in elections, we prefer polls of likely voters2 over registered voters over all adults. For presidential approval, we use polls of all adults over registered voters over likely voters. This is done for the sake of historic comparison, since approval has been measured since the 1930s among adults.

  • If a pollster has multiple likely voter screens, they are averaged together without preference.

• Leaners: we prefer polls that push leaners (essentially the version with the fewest undecideds).

• Minor third party candidates: which we define as candidates not consistently polling at 10% or higher. We use the version with the fewest minor candidates until October. That’s because minor candidates tend to be overestimated early on and can affect accuracy.

• Wording: if a poll informs respondents of information beyond what will be on the ballot, it will not be included.

  • The generic ballot average is a bit complicated because it doesn’t measure an actual vote. Instead, we include two types of wording: which party’s candidate do you want for Congress and which party do you want to control Congress. If a pollster asks both questions, we use the candidate one. Vaguer questions are excluded (some pollsters throw in regardless of how you’ll vote or will it be better for the country if one party controls Congress).

2. Weighting

We assign each poll five weights:

Recency: as polls age, their weight decays exponentially. For election averages, weighting is less aggressive early on. This is done by calculating our averages on two settings—one slow, one fast—and gradually shifting to the fast one. For averages where less than three polls are available for over six months, we default to the slow setting to avoid giving any one poll too much weight.

Sample size: larger samples are favored in all of our averages. We use the same system as the now defunct 538 averages, weighting polls by the square root of their sample size3, with no additional weight for very large polls (1500 for election averages, 10000 for approval).

Population polled: elections are determined by those most likely to vote, so our state/district-level averages favor likely and registered voter polls (weighted 1 and 0.8, respectively), followed by adult polls (0.4). National polling (generic ballot and approval) tends to have a lot more data on the difference between populations, so we adjust polls towards the preferred population instead of weighting.

Pollster quality: all of our current and recent averages are weighted using Silver Bulletin ratings, while historic averages use the old 538 version of those ratings.4 Pollster weights decline 5% with each grade drop (A+ weighted 1, A 0.95, A- 0.9, and so on). Ungraded pollsters get a default grade of 0.5.

Lastly, we limit the weight of pollsters that release surveys frequently to prevent any one firm from dominating the average:

weight = 1 / polls released in 3-week window

All five weights are then multiplied together to determine a poll's overall weight.

3. Initial adjustments

A few factors can tell if us a poll is “biased”—not in terms of accuracy, since we can’t know that ahead of time—but in terms of consistently deviating from other polls:

1. Population polled: whether a poll uses an adult or registered/likely voter screen. To handle polls of different populations, we look at pollsters who release results for different populations in the same poll. Polls are then adjusted towards likely voters for our generic ballot average, and all adults for presidential approval.

2. National adjustment: polls at the state/district level are adjusted for shifts in the national environment (as measured by the generic ballot in congressional contests). This is based on the trend line adjustment pioneered by 538.

4. House effects adjustments

Pollsters have “house effects,” or something in their methodology that produces consistently different results from other firms. To address this, our model learns each pollster’s bias. This is done by tracking how a pollster deviates from the weighted and adjusted average of all other pollsters, as explained in steps 1-3.

The house effects adjustment gets stronger if we have more polls from a firm, and have many other firms to compare them to. For state-level polling, we estimate house effects using data from different states and national polling for more robust estimates.

5. Final average

The final average (the one you see) aggregates the adjusted results, with slightly less aggressive weighting to avoid double-counting. We also drop extreme outliers at this stage in our election averages. Polls that are completely dropped must qualify in three ways: not polling likely voters, having a high undecided vote share, and deviating from other pollsters by twice the historic average. Polls with extremely high undecideds (at least 15% above other pollsters) are included but do not affect the undecided vote share.

Check out some of our averages below!

NEW: Senate and House forecast

Polling averages

Governor / Senate / Data centers / Presidential approval / Generic ballot

Methodology

Live polling average

Senate forecast / House forecast / Polls

What’s new?

Democrats are favored to win the Senate. Republicans aren’t out of the fight, but they do need a consistent polling error in their favor across multiple battleground states. Should the election go as expected, Democrats will win somewhere around 52 Senate seats—a downright herculean feat in theory. They’ll have to hold two swing states, flip North Carolina for the first time since 2008, unseat a five-term incumbent in Maine, and win three states that voted for Trump by double digits in 2024, including Texas, the crown jewel of the Republican Party.

We will be live blogging and analyzing the election as it happens here. Check out our county benchmarks, which will be updated with actual results as they come in, here. You can also read our hour-by-hour guide to the election here.

Bonus features

Subscribe now

Subscribe for free to explore additional data and interactive charts (updated daily), no Substack account needed. Free subscribers can access:

• The data used to make each state’s forecast

• Which states are most likely to decide Senate control

• Each party’s seats and odds forecast over time

• How winning certain states affects each party’s odds

• What this model would have predicted for past Senate elections

• Articles and similar bonus data for our House forecast

Senate forecast / House forecast / Polls

What’s new?

Barring a major last-minute scandal or poll drop, our forecast is final: Democrats have a greater than 95% chance of flipping the House. A Republican hold, while technically possible, would be a historic upset. Alternatively, if the polls are broadly correct, Democrats will win their biggest House majority since 2008.

Bonus features

Subscribe now

Subscribe for free to explore additional data and interactive charts (updated daily), no Substack account needed. Free subscribers can access:

• The data used to make each district’s forecast

• Each party’s seats and odds forecast over time

• What this model would have predicted for past House elections

• Qualitative ratings and uncertainty

• Articles and similar bonus data for our Senate forecast

Graham Platner, the Democratic nominee for Maine’s Senate race, is in trouble. On Monday, after Politico published rape allegations against the candidate, multiple former supporters called on him to drop out, including Congressman Ro Khanna and Senators Bernie Sanders and Elizabeth Warren.

The electoral calculus is clear: scandal-plagued Senate candidates have historically underperformed. But what if Platner heeds calls to drop out? Maine Democrats can name a new nominee if he quits by July 13, and the race to replace him is underway. Troy Jackson, the former Maine Senate president, has filed paperwork to run, and former Maine CDC director Nirav Shah and Secretary of State Shanna Bellows have expressed interest. But what’s the precedent for such a candidate swap?

Share

It’s not uncommon for House and Senate incumbents to resign over allegations of wrongdoing, but they almost always do so before winning a primary. To find a remotely comparable Senate nominee, we have to go back to 2014, when Montana Senator John Walsh withdrew over allegations of plagiarism. His replacement, Democratic state representative Amanda Curtis, polled worse than he did and lost the general election by nearly 18%.

While there was a polling collapse for Democrats after Walsh dropped out, his case is hard to draw conclusions from. He himself never took a visible polling hit from the scandal, and his successor’s defeat could be chalked up to being a non-incumbent Democrat in a red state (not to mention that 2014 was a tough year for Democrats nationally).

Before Walsh, there was the 2004 Illinois Senate race. Months after winning the GOP primary, Republican nominee Jack Ryan dropped out over reports that he had pushed his wife to perform public sexual acts. Pre-scandal polling was limited, but Ryan trailed Democrat Barack Obama by an average of 15% in two polls. His replacement, former diplomat Alan Keyes, later trailed Obama by a far worse margin of 40% for the rest of the campaign, and lost by roughly that much.

While this was a disastrous showing for Republicans by Illinois standards, Keyes himself was a perennial candidate from Maryland. In the end, his underfunded campaign was no match for Obama’s.

The other example of a last-minute swap this century came out of New Jersey in 2002. Incumbent Robert Torricelli was sinking amidst an ethics scandal, forcing Democrats to replace him with former Senator Frank Lautenberg. This arguably provides a cleaner case study than Montana or Illinois, since Democrats put up a strong replacement, and the race was heavily polled.

In the end, Lautenberg won by just under 10%, roughly matching New Jersey’s partisan lean at the time. However, for a former Senator, this was far from an impressive victory, especially since his Republican opponent was but a small town mayor. The polls (which were relatively accurate in the race) showed Lautenberg failing to recover Torricelli’s pre-scandal lead.

For a few more recent examples, we can look to House races. The exact level of underperformance varied widely across races, but most replacement House candidates did not do well.

If the Democratic Party is to replace Platner, history suggests that a new candidate may not be the end-all, be-all. This makes sense since any nominee faces several hurdles: avoid being tainted by the scandal, enter a race with less than four months to go, and win over primary voters who backed Platner a month ago.

What if Platner doesn’t drop out? In the (unlikely) scenario where he stays in the race, Maine Democrats can mount a risky write-in campaign. Only two Senators have ever won this way: Strom Thurmond in 1954, and Lisa Murkowski in 2010. The former was the sitting governor and beat a single candidate from his own party. Murkowski beat both the Democrat and the Republican in her race, but she was the incumbent, and the Republican vote was split nearly 50-50.

The most relevant campaign might be Nebraska’s 2020 Senate race. Democrats were forced to run a write-in after their nominee, Chris Janicek, refused to drop out over a sexual comments scandal. The write-in Democrat won just 6% of the vote to Janicek’s 24%. Even if you combined the two Democrats’ votes, they got far less than the 39% Joe Biden received on the same ticket—a degree of underperformance Maine Democrats cannot afford this year.

NEW: Senate and House forecast

Other polling averages

Governor / Senate / Data centers / Presidential approval / Generic ballot

Methodology

General election polling average

Learn more about how these averages work here.

On May 26, Texas Attorney General Ken Paxton trounced Senator John Cornyn in the state’s Senate primary election. Within an hour, Cook Political Report and Crystal Ball downgraded their general election forecasts from likely to lean Republican. After all, the downsides of nominating Paxton were no secret: allegations of securities fraud, bribery, and retaliation, impeachment by his own party, and divorce on “biblical grounds.” But how much do scandals really affect Senate races?

Share

Since 2006, at least 23 Senate nominees have experienced a pre-election scandal of some kind, excluding off-cycle special elections.1 I’ve added a wide range of allegations to the list below, including any formal investigations and actual crimes. The list errs on the side of caution when it comes to controversial statements.2

As you can see, the average affected candidate did over 6% worse than expected, given their state’s partisan lean and the national environment. Only five candidates didn’t underperform, all but one of whom had the incumbency advantage. One also faced an inexperienced opponent with his own (arguably far worse) scandals.

Meanwhile, the two candidates with corruption-related allegations like Paxton—Bob Menendez and Ted Stevens—posted especially large underperformances of 8.5% and 19.2%. Both were incumbents.

Under the same baseline, the 2026 Texas race starts at R+5: the state has recently voted about 12% more Republican than the nation as a whole, but Republicans are also down over 7% in national polls. For context, the 2024 baseline would have been around R+13, as Texas had voted about 11% more Republican up to that point, and congressional Republicans won by just over 2% nationally. In other words, due to the Democratic-leaning environment this year, Paxton would be expected to lose with a scandal penalty exceeding 5%.

However, scandals don’t seem to impact elections as much as they used to, probably due to high political polarization. The recent candidates who did significantly worse than baseline (Royce White, Sarah Gideon, and Bernie Moreno) also faced strong opponents. But that doesn’t mean scandals are meaningless—just last November, a scandal-plagued Democrat significantly underperformed other Democrats on the same ticket in Virginia’s elections.

A more complicated regression model, accounting for rising polarization3, candidate experience4, and the incumbent’s last election margin relative to baseline, finds an average penalty of about 3% under current levels of polarization.

Even with this refined estimate, it’s important to remember that no two scandals are the same. As such, we can’t simply project the historic average onto the Texas Senate race—though it’s hard to argue that Paxton’s scandals will be any less severe than those of past candidates.

For one, he’s faced multiple allegations spanning infidelity and corruption (potentially more toxic to voters than the average allegation, as previously discussed). This contrasts with most recent nominees, who tended to have one issue. Voter awareness will also be high—Paxton’s baggage is expected to be the focal point of Democratic messaging, not to mention the tens of millions in ad spending already released by his Republican primary opponent.

In fact, in a recent poll asking non-Paxton voters why they weren’t backing the attorney general, many responses contained words like “crook” and “criminal.” This is a salient issue. Patrick Ruffini, who runs Echelon Insights, one of the most accurate Republican polling firms, summarizes the situation quite well:

Ken Paxton is somewhere between Herschel Walker and Roy Moore in candidate quality—and a lot hinges on where he lands on that spectrum. But the campaign to polarize the conservative electorate against James Talarico—with comments heavy on veganism and gender ideology—has only begun. And in a state like Texas, polarization is usually enough.

Importantly, if Ruffini is right about Paxton being on the level of Walker and Moore5, polarization and Texas’s partisan lean probably won’t be enough to carry Republicans over the finish line. The former needed a nearly R+8 baseline to win, and the latter (who is harder to evaluate because he ran in an off-cycle race) needed something on the order of R+20.

Even if Paxton is not as damaged as Moore and Walker, something closer to the average 3% scandal penalty can absolutely make the difference—the last Texas Senate race in a good year for Democrats (2018) was decided by less than 2.6 points. If anything, national polling currently looks better for Democrats than it did in 2018. Plenty could change by Election Day, but for now, blue Texas looks more plausible than it did in any recent election.

Polling average

Donald Trump has never been so unpopular. Just 38.4% of Americans approve of his job performance in our average of recent polls, compared to 59.1% who disapprove. His net approval is -20.7%, down from -12.9% at the start of the year, -20.1% after the events of January 6, and lower than his all-time low in 2017, or Joe Biden’s worst numbers after the 2024 presidential debate.1

Share

These numbers mark a dramatic shift from when Trump first returned to the White House last year. He maintained positive net approval through mid-March 2025, and held steady at -4 to -9% in the summer and fall—much better than the -14 to -20% ratings recorded at the same point of his first term.

Unlike during his first term, however, Trump is not experiencing a year two rebound, driven in 2018 by his strong economic approval. Looking at the issues, it’s not hard to see why. His economic ratings currently sit about 28% underwater, a far cry from the positive net approval recorded in nearly every poll of his first term. Regarding inflation, less than 27% of Americans approve, compared to 69% who disapprove. These are disastrous numbers considering that inflation and the economy are polling as Americans’ top two issues.

Furthermore, just like his gains with voters leading up to the 2024 election, Trump’s popularity collapse is broad. His net approval is down significantly with every group in our average, but it’s especially poor among independents, who currently sit at -40.9%. The only bright spot comes from Trump’s own party—Republicans approve of him by over 60%—though even that figure represents a double-digit drop from early 2025.

Three main events appear to have contributed most to these numbers: the Liberation Day tariffs, the government shutdown, and the Iran War. While all three events look similar in magnitude—each causing 5-6% drops in net approval—the Iran War stands out for two reasons. For one, the president was already at a low point before the war, with limited soft approval left to lose. He’s also still bleeding support more than two months in, compared to the month-long shutdown and tariff downturns.

After all, the war isn’t over. A ceasefire is in place, but the flow of oil through the Strait of Hormuz is still being disrupted, and gas prices remain elevated from before the war. Inflation is at a rate not seen since 2023, rising in lockstep with the president’s falling popularity.

With that being said, Trump is not facing reelection in 2026. That might alleviate the pressure of public opinion to an extent, but his party still has to survive the midterm elections this fall. Our average of generic ballot polls, asking voters which party they intend to vote for, has closely tracked Trump’s approval. Democrats are now up by almost 7%, suggesting that the GOP’s fortunes remain tied to the president’s sinking popularity.

For context, 7% represents a larger advantage than the 5% lead enjoyed by the GOP in the 2010 and 2014 midterms, and slightly short of Democrats’ lead in 2006 and 2018 (a little over 7% ).2 Under those kinds of conditions, Republicans will almost certainly struggle to maintain their majorities in Congress. Of course, there are almost six months to go before voters cast their ballots. The president and his party still have time to turn things around—or collapse to new lows.

If the polls are to be believed, Democrats are on track for a smooth election night this November. A year into his second term, Donald Trump’s approval is at historic lows, and Republicans are down 6% in national polls. That puts Democrats in a strong position to flip the House, and all eyes are now on the Senate, where Republicans have trailed in polls of Maine and North Carolina, not to mention the upset results we’re seeing out of red states like Texas and Alaska.

Share

But can these polls be trusted? To investigate, I collected and averaged polls of every two-party November Senate race I could find going back to 2006, totaling nearly 8000 individual surveys across over 300 races.1 On average, the final polling aggregate missed the margin of victory by 4.7%. Root mean square error (RMSE), which is more sensitive to major misses, was 6.3%.

Electoral politics obviously aren’t stationary, however, and we’re still half a year out from November—plenty of time for voters to change their minds. Zoom out to 6 months before Election Day, and average error rises to 7%, while RMSE reaches 8.8%. Even so, early polls do correlate pretty strongly with actual results:

In practice, a 7% miss on the margins is far from trivial. If we shifted every competitive Senate race this year 7% to the left, Democrats would comfortably retake the Senate. Red states like Ohio, Texas, and Florida would flip, and even South Carolina would come into play.

An equal miss in the opposite direction would expand the already substantial Republican Senate majority. That means Democratic losses in Georgia, New Hampshire, and Minnesota, likely enough to guarantee a GOP majority through 2030. These may seem like totally different scenarios, yet both are technically plausible from a polling standpoint.

Another problem with early polling bias is its unpredictability. Across all Senate races, polling error at this point overestimated Democrats by as much as 6% in 2010 and Republicans by 4% in 2012. Bias was similarly volatile in the most competitive states, where Republicans beat their margins by 8% in 2014, and Democrats did so by 5% in 2008.

When can we expect more accurate data? Historically, voter sentiments have shifted most dramatically in October, about 1-2% on the margins each week. Movement is comparatively sluggish early in the cycle, meaning a survey result from August isn’t all that different from one in May. Until then, take the polls with a heavy grain of salt—but don’t ignore them entirely.

Election night was a doozy. In Wisconsin, a state where each of the last three presidential elections was decided by less than a point, the Supreme Court race proved to be anything but close: liberal Chris Taylor beat her conservative opponent, Maria Lazar, by around 20%. Depending on where the final votes settle, Taylor’s victory will be the largest of any court race since 2009 and possibly 2000, both of which involved incumbent justices.1

Lazar’s defeat came on the heels of the far more prominent 2025 court election, when billionaire Elon Musk came to the aid of the Wisconsin GOP (though they still lost by 10%). This time around, spending was far more limited, and Taylor out-raised Lazar by more than 6-to-1.

With little outside help, Lazar failed to mount a strong showing in the crucial suburban “WOW” counties of Waukesha, Washington, and Ozaukee (which Taylor flipped). Her performance likely helped nudge Democrat Alicia Halvensleben over the finish line in the mayoral race in Waukesha, a city President Trump carried by almost 7% in 2024.

Wisconsin’s Supreme Court will now have five liberal and two conservative justices, a dramatic shift from just three years ago, when Democrats flipped the court for the first time in 15 years. Their majority is safe through 2030, which will prove vital as redistricting heats up in other states and the next census approaches, bringing legally contentious questions around redistricting.

Down in Georgia’s 14th Congressional District, Democrats mounted an equally impressive performance, even if they fell short of outright victory. Their nominee, Shawn Harris, lost by just 12%, 16 points closer to victory than himself in 2024, 25 better than Kamala Harris, and 26 better than Joe Biden in 2020. In fact, GA-14 saw the best Democratic performance out of any special House election since Trump’s inauguration, adding to a cycle already full of historic shifts.

These results are undeniably part of a broader trend. Boosted by their increasingly educated and politically engaged coalition, as well as heightened enthusiasm against Trump, Democrats are doing better in off-cycle races where turnout is low. In Wisconsin, turnout was down by more than half compared to 2024, while in GA-14, it was down nearly two-thirds.

The turnout regression does predict much smaller Democratic overperformances in higher-turnout races, so the midterm environment is unlikely to shift the way Wisconsin and Georgia shifted last night. Even so, midterm elections tend to see fewer votes cast and electorates skewed against the president’s party, meaning the GOP remains in hot water for 2026.

Additionally, both of Tuesday’s races took place in critical swing states. In Georgia, Republicans are hoping to hold the governorship and retake one of the state’s Senate seats to protect their majority. In Wisconsin, another governorship is wide open, and two GOP House incumbents face competitive reelection bids. If the Republican candidates in those races were hoping for good news last night, they were sorely disappointed.