Who controls Congress after November?
Win probabilities for every competitive 2026 race, from a model trained on 25 years of polls and results — and graded in public against everyone else who calls elections.
Senate →
100 seats · 51 to winDemocrats end up with 49 seats on average — 46 to 52 in four scenarios out of five. They need 51.
House →
435 seats · 218 to winDemocrats end up with 215 seats on average — 205 to 225 in four scenarios out of five. They need 218.
Governor →
36 states · 23 forecast7 look safe or likely Democratic and 2 safe or likely Republican, with 3 genuine toss-ups. The remaining 13 states on the ballot aren't close enough to forecast.
How the forecast has moved
Each chamber's control probability, with the model rerun as it would have read on every date — the Senate from the race polls available then, the House from the national generic ballot.
Senate
D +25 since JanuaryHouse
R +21 since JanuaryRaces that will decide it
all 46 races →The six closest contests. Anything inside ten points can go either way.
On the campaign trail
candidates & the courts
How to read these numbers
A probability is not a prediction. When this model says 70%, it means the race looks like ones where that side won about seven times in ten — and lost the other three. Those three happen. The probabilities are calibrated, which is a claim you can check: every past forecast is scored on the track-record board. Read the methodology →