Explainer · Election-week series · Nov 4, 2026 (results still being counted; uncalled races noted)
Control of both chambers is called; final seat counts are not yet determined. Do not present the projected House total or the Alaska Senate result as final. Results are unofficial until states certify them.
On the eve of the election, forecasts pointed one way: Democrats were favored to take the House comfortably and, with wide disagreement, narrowly to take the Senate. Voters have now had their say. Here is what they decided, and how those forecasts held up.
The Senate: Democrats take control
What was forecast. Decision Desk HQ gave Democrats a 63% chance of winning the Senate, projecting 52 seats. Nate Silver’s model put it at 79%. The Economist was the skeptic at 53%. Prediction markets sat near 68%. The spread from 53% to 79% was the story: every forecaster leaned the same direction, but none agreed on how far. CNN’s range of plausible outcomes ran from 49 to 55 Democratic seats. Cook rated seven races as toss-ups — six Republican-held (Alaska, Iowa, Kansas, Maine, Ohio, Texas) and one Democratic-held (Michigan) — and estimated a Democratic gain of 2 to 6 seats. The Democratic path, as laid out before the election, was to hold Michigan, flip North Carolina, and win three of the six Republican-held toss-ups.
What happened. Democrats won Senate control. The Associated Press called Senate control at 11:37 p.m. ET on Nov. 3. Democrats flipped seven seats so far, a net gain of seven, taking the Democratic caucus from 47 to at least 54. This beats Decision Desk HQ’s 52-seat projection and sits just above Cook’s +2 to +6 gain range. Alaska, the last Cook toss-up, was still uncalled on the morning of Nov. 4, so the final total is 54 if Republican Dan Sullivan holds it, 55 if Democrat Mary Peltola wins it. Mississippi was also still uncalled in the AP tally.
The toss-ups, one by one.
- Michigan (Democratic-held, must-hold): held — Abdul El-Sayed (D) beat Mike Rogers (R).
- North Carolina: flipped — Roy Cooper (D) beat Michael Whatley (R), about 58.7% to 39.1% with 99% counted (AP/Fox call).
- Alaska: uncalled — Mary Peltola (D) vs. Sen. Dan Sullivan (R).
- Iowa: flipped — Josh Turek (D) beat Rep. Ashley Hinson (R), open seat.
- Kansas: flipped — Adam Hamilton (D) beat Sen. Roger Marshall (R).
- Maine: flipped — Troy Jackson (D) beat Sen. Susan Collins (R); Collins conceded.
- Ohio (special): flipped — Sherrod Brown (D) beat Sen. Jon Husted (R).
- Texas: flipped — James Talarico (D) beat Ken Paxton (R), open seat.
- Beyond the toss-ups, Democrats also flipped South Carolina: Annie Andrews (D) beat Sen. Darline Graham (R). Democrats held Georgia (Jon Ossoff), Virginia (Mark Warner) and New Hampshire (Chris Pappas).
Closest forecaster. On direction, all of them: every Senate forecaster leaned Democratic, and Democrats won. On seat count, Decision Desk HQ’s 52 understates the called result of at least 54; Cook’s +2 to +6 gain range also understates a +7 gain so far. Silver’s 79% and the markets’ ~68% were the most confident Democratic calls, and the outcome landed on their side of the spread. But a probability is not a seat count, and the miss to note is that the forecasters’ central seat estimates were low, not that their direction was wrong.The Economist’s 53% was the most cautious framing, and it too pointed the right way.
The House: Democrats take control
What was forecast. Decision Desk HQ gave Democrats an 83% chance, projecting 241 seats. Prediction markets were higher, near 92–93%. Cook estimated a Democratic gain of 12 to 22 seats. The generic ballot averaged D+8.3 across 45 polls. Democrats needed a net gain of 4 seats to reach 218, from a starting point of 218 Republicans, 214 Democrats, one independent caucusing with Republicans, and two vacancies.
What happened. Democrats won House control. The Associated Press called House control at 1:28 a.m. ET on Nov. 4; NBC News, CBS News and CNN also projected a Democratic House. The AP had called 223 Democratic seats by the morning of Nov. 4, with 81 seats still uncalled in one tally and Democrats leading by 41 more in the AP’s district count. So 223 is a floor, not a final total.
NBC News projected Democrats will win at least 244 seats. That projected floor already clears Decision Desk HQ’s 241 and, against the 214 seated Democrats going in, implies a gain above Cook’s +12 to +22 range. But the final count is still moving, so do not publish a final House number yet.
Closest forecaster. On control, all of them—DDHQ at 83% and the markets at ~92–93%—pointed to a Democratic House, and that is what happened. On magnitude, the early projection of at least 24) runs ahead of DDHQ’s 241 and above Cook’s gain range. The generic ballot’s D+8.3 pointed to a comfortable, not narrow, win, and that is the shape of the result so far.
Governors: the four toss-ups
What was forecast. Cook’s final ratings called four toss-up governors’ races: Wisconsin, Georgia, Nevada, and Kansas — with Ohio rated Lean Democratic, and the broader map at 13 Solid Democratic, 6 Lean Democratic, 3 Lean Republican, 3 Likely Republican, and 7 Solid Republican.
What happened.
- Wisconsin: held for Democrats — David Crowley (D) beat Tom Tiffany (R).
- Georgia: flipped — Keisha Lance Bottoms (D) won the open seat.
- Nevada: flipped — Aaron Ford (D) beat incumbent Gov. Joe Lombardo (R).
- Kansas: still uncalled in the AP tally on the morning of Nov. 4. Democratic-held open seat; Ty Masterson (R) led in one tracker, but the race was not called. Do not call it in this piece.
Beyond the toss-ups, the surprises were Texas — Gina Hinojosa (D) beat Gov. Greg Abbott (R), the first Democratic Texas governor since 1990 — Ohio — Amy Acton (D) beat Vivek Ramaswamy (R), in line with its Lean Democratic rating — and Iowa — Rob Sand (D) won the open seat.
Ballot measures: what voters settled directly
What was forecast. 145 statewide measures were on ballots in 39 states, including 39 election-related measures — voter ID, redistricting, and primary rules among them.
What happened (reported results, unofficial until certified). Voter ID split by route to the ballot: California and Colorado voters defeated citizen-initiated voter-ID measures. Colorado’s mail-ballot ID amendment, Amendment 84, trailed about 36% to 64%. Ohio voters approved a legislature-referred measure that would put enhanced voter-ID requirements in the state constitution.
Nevada’s voter-ID amendment, already approved once in 2024, reportedly passed its required second vote. On redistricting: Maryland voters approved an amendment that could allow new congressional maps ahead of 2028, while Missouri voters rejected Republican-drawn House districts that courts had blocked unless voters approved them. Colorado’s redistricting amendment (Amendment 86) trailed about 46% to 54%.
On abortion: Virginia voters approved a constitutional amendment creating a fundamental right to reproductive freedom through the first two trimesters, and Missouri voters rejected a proposed amendment that would have reimposed abortion restrictions. Do not present a complete state-by-state measure list — cite only these verified measures, as in the Nov 1 piece.
What it means
With Democrats controlling both chambers and a Republican president, the next two years are divided government. In practical terms, the Democratic Senate can block presidential nominees, including any Supreme Court vacancy. Democratic majorities in both chambers can pass bills, but they may still face a presidential veto. So the legislative agenda, confirmations, investigations, and the 2028 landscape will all run through negotiation or stalemate, not one-party action.
The new Congress is seated January 3, 2027; until then, the outgoing Congress sits in a lame-duck session. That arithmetic — what each side can now pass, block, or must negotiate — is the story to watch, not a single race. forecast scorecard
| Decision Desk HQ | 63% / 52 seats | Control D; at least 54 seats, Alaska uncalled (52 understates it) | 83% / 241 seats | Control D; NBC projects at least 244 (final count pending) |
| Nate Silver | 79% | Direction correct | — | — |
| The Economist | 53% | Direction correct, most cautious | — | — |
| Prediction markets | ~68% | Direction correct | ~92–93% | Direction correct |
| Cook | +2–6 seats | +7 so far (above range) | +12–22 seats | Gain tracking above range on early projection; final count pending |
A probability is not a prediction, and this election is the receipt. A 63% forecast that comes true still leaves a 37% chance of the other outcome. A 79% forecast that fails is not “wrong” the way a missed score is wrong. But the forecaster whose seat count lands closest earns the most trust next time. That is the standard these numbers should be judged by, and the one we will use when the next cycle’s forecasts arrive.
Races still uncalled at publication time are noted as such: Alaska Senate, Mississippi Senate, Kansas governor, and the remaining House seats. Results are unofficial until states certify them.