The final FiftyPlusOne election forecast model for the 2026 U.S. congressional midterms gives Democrats a 79 percent chance of winning the majority of both chambers of the U.S. Congress today. The party has a 99 percent chance of taking back the House and a 79 percent chance in the Senate.
Now, 79 percent is not 100. Democrats do lead in seven U.S. Senate seats that are currently held by Republicans. But as I explained here, leads in four of those states are small enough that a polling error similar to the size America saw in 2020 or 2024 would be enough to cut Democrats’ gains to just 50 seats, up from 47 today. Democrats need 51 seats to break the vice president’s tie-breaking vote in the chamber.
It is worth noting that Democrats have also been gaining ground in the home stretch. In eight of the closest 12 Senate races, Democrats have gained in the polls over their Republican competitors over the last two weeks. This has pushed them out of the “Toss-up” category and into “Lean Democratic” territory in several key seats, per our forecast.
After the decade pollsters have had, some skepticism of these pre-election surveys is justifiable. But they aren’t the only indicator of a strong environment for Democrats.
The nonpartisan race raters, whose judgments draw on far more than polls, have been moving in the same direction. Over the past two weeks, the Cook Political Report, Inside Elections, and Sabato’s Crystal Ball at the University of Virginia Center for Politics have shifted dozens of House and Senate races, and almost all of those changes have favored Democrats.
That matters because the raters see things polls often don’t. They look at candidate quality and fundraising, and they read the private and internal polling that campaigns and party committees never publish. They also spend their days talking to operatives and local reporters about which way each race is moving. And they rate every seat, including the many House districts that get one or no public polls all cycle. The raters aren’t fully independent of the polls (they read them too), but when both point the same way, we can be more confident the signal is real.
With a little bit of work, FiftyPlusOne has translated individual seat-level ratings from these major handicappers into aggregate chamber-level forecasts for the House and Senate.
Paid subscribers to 50+1 get access to premium analysis, plus sortable tables and complete data access on our polling website. If you want to follow the 2026 cycle with the best data at your fingertips, become a paid subscriber.
Turning ratings into odds
A rating is a label, not a probability, so the first step is translating one into the other. Rather than assume “Lean Republican” means some fixed number, I looked at how each rater’s labels have actually performed. For each of the three raters, separately for the House and Senate, I collected their final pre-election ratings in every midterm from 2010 to 2022. I then counted how often the president’s party won in each category.
That history matters, because in midterms the rating buckets haven’t been symmetric. In Cook’s House ratings, for example:
Seats leaning toward the president’s party went his way about 86 percent of the time.
Seats leaning toward the opposition went his way less than 2 percent of the time.
Toss-ups broke for the president’s party only about 40 percent of the time.
Midterm toss-ups tend to break against the White House, and the other two raters show the same pattern. This year, the president’s party is the Republicans.
Seats also don’t move independently. If the national environment is a few points better for one party than expected, it shows up across the map. So in each simulation, I add a shared national error that shifts every competitive race together. Its size comes from how far past midterms’ results landed from the final ratings.
I then simulate the election 20,000 times for each rater’s latest map: Cook’s and Sabato’s from Nov. 2, and Inside Elections’ from Oct. 29, the most recent data available at the time of writing. Each simulation produces a seat count for each chamber. The share of simulations where Democrats reach 218 House seats, or 51 Senate seats, is that rater’s implied chance of a majority.
What the ratings say today
The chart below shows the distribution of simulated seat totals for each rater. This is the distribution for the House:
And the Senate:
In the House, all three raters’ maps imply an essentially certain Democratic majority. The median outcome is 235 seats under Cook, 236 under Inside Elections, and 241 under Sabato. That matches our forecast’s 99 percent.
The Senate is where they disagree:
Cook is the most cautious, mostly because it still rates six Senate races as Toss-ups: Alaska, Iowa, Kansas, Maine, Ohio, and Texas. Sabato’s final map has none. Averaging the three gives Democrats about a 76 percent chance of winning the Senate, with a median of 52 seats. That’s within a few points of our forecast’s 80 percent.
History is doing a lot of work here. If you took the labels at face value, treating a Toss-up as 50-50 and a Lean race as 75-25, the ratings would give Democrats only about a 62 percent chance in the Senate and 84 percent in the House. The difference is the midterm penalty the president’s party has paid in the past, which the labels alone don’t capture.
Race raters have been playing catch-up with quantitative forecasts like FiftyPlusOne
Finally, I ran the same exercise for every day since June. Each day uses each rater’s ratings as of that date, compared with their historical record at the same distance from Election Day.
Here’s each rater’s implied probability of winning the U.S. House this election cycle, per our modeling of their ratings:
And the Senate:
Two things stand out. First, the raters have moved a lot. In June, their ratings implied about a two-in-three chance of a Democratic House and roughly a one-in-three chance of a Democratic Senate. Since then, the House has climbed to near certainty and the Senate to roughly three-in-four. Much of that movement came in steps, as the raters shifted batches of races at once in mid-July and again through October. Cook’s final update, which moved Michigan from Toss-up to Lean Democratic, added about 10 points to its Senate odds on its own.
Second, our model (the dashed line) was ahead of the raters for most of the year. Through the summer and early fall, it gave Democrats better odds in both chambers than the ratings did, and the ratings have spent the last two months catching up. Today the House gap is gone, and the Senate gap is down to a few points, mostly reflecting Cook’s more cautious map.
Our final forecast
So here’s where things stand the morning of the election. Our final forecast gives Democrats a 99 percent chance of winning the House, a 79 percent chance of winning the Senate and a 79 percent chance of winning both. The race raters, reading the races through their own reporting and their own track records, end up in nearly the same place: an all-but-certain Democratic House and roughly a three-in-four chance in the Senate.
That agreement is reassuring, but it isn’t a guarantee of anything. Most of the raters’ remaining caution sits in the same handful of Senate races where the polls are closest. If the polls miss tomorrow the way they did in 2020 or 2024, a Democratic majority in the Senate would be far from guaranteed. Our forecast would view Republicans holding the Senate as a surprise, but not as a complete shock.
That’s the best way to interpret a roughly 8-in-10 chance: It’s not a prediction that Democrats will win, but a statement of how often parties have won past elections when they were in positions similar to the one Democrats find themselves in today. Most of the time, the evidence we have tonight says, Democrats take both chambers. Some of the time, they don’t.
This is our last update. The forecast is now locked. Voters will get the final say shortly.
Please join FiftyPlusOne’s Mary Radcliffe and G. Elliott Morris for an election night livestream with the teams from The Downballot and Strength In Numbers at 6:45 p.m. ET, right here on the Strength In Numbers website or via the Substack app. We know there are a lot of places you could be on election night, but we think our joint offering will be the best combination of entertainment and information that you can get. It will also be entirely free of commercials. There will be a lot of caffeine and leftover Halloween candy.
And you can click below to save a reminder to your calendar:
Apple | Google | Office 365 | Outlook | Outlook.com | Yahoo