Will Democrats win control of the U.S. House in the 2026 midterm elections?

Midterms are referendums on the party in the White House, and the October readings show no recovery for it. A Fox News poll reported on October 14 put the president's approval at 36 percent with 64 percent disapproving, a low for that survey, and an Associated Press/NORC poll reported a day later found the same split. Voters unhappy with prices and the Iran war have one lever in November, the House ballot.

The most telling behavior is what Republican groups now do with money rather than what they say. Speaker Mike Johnson still tells reporters the polls are wrong, but the Congressional Leadership Fund and Trump-aligned groups began pulling support from a pair of endangered Republican incumbents on October 13, and a day later the same super PAC dropped its effort to defeat Democratic Rep. Henry Cuellar in a Texas district Republicans redrew in their favor. Committees cut seats only when their own private polling says the seat is gone; the Cuellar retreat is a lost Republican pickup chance rather than a new Democratic gain, but it shrinks the cushion Republicans were counting on from the new maps.

The reference class is every midterm since 1946. The president's party lost House seats in 18 of them, per The Conversation's count, and the two exceptions, 1998 and 2002, came under presidents with high approval. In the American Presidency Project table, every postwar midterm held under a president approved by less than half the public cost his party House seats.

The closest analogue is still 2018, Trump's first midterm, when Democrats won the House with a generic ballot lead close to today's. The analogy breaks on the map: mid-decade redistricting nets Republicans about 12 seats against 1 for Democrats on Wikipedia's tally of the new maps. Cook's October update suggests the national environment is now large enough to swamp that structural edge, since its reported projection puts the Democratic gain well beyond the three seats needed.

Our October 12 number was 91 percent, built from a 90 percent base rate of midterm seat losses (18 of 20 since 1946), adjusted up for the environment and Cook's October shifts and down for the new maps. Since then our up signpost, the Congressional Leadership Fund cancelling support in Republican-held Toss Ups, has fired: Punchbowl News reported on October 13 that the group and Trump-aligned committees began to abandon a pair of endangered Republican incumbents, and on October 14 that it had given up on the Cuellar seat. The down signpost did not fire: a NewsNation and Decision Desk HQ poll reported on October 13 had Democrats ahead on the generic ballot 49 to 42, well above the D+5 trigger.

Arithmetic, our calculation: 91 plus 3 for the signpost is 94. We take back 2 points because the retreat mostly reflects the same polling we had already counted, and because the Cuellar seat is Democratic-held, so dropping it removes a Republican offensive target rather than adding a Democratic one. That lands at 92, two points above the Kalshi price. We keep 8 points for a Republican hold because Republican-drawn maps still protect several seats, and a polling miss of the 2022 kind, which Politico reported on October 15 is now Republicans' main hope, would shrink a projected wave toward a narrow result.