Houston Area 2026 forecast

How this works

This is a thought exercise as much as a forecast: change the assumptions above and see what it would take to change an outcome. It is a projection from early-vote rosters and past results, not a poll, and it will be wrong in places.

Who is voting. Every 2026 ballot cast so far is on the county's early-vote and mail rosters. Each voter is grouped by their most recent March primary on file (2022, 2024 or 2026): Democratic, Republican, or none. The rest of the electorate is projected precinct by precinct from how many more ballots each group cast after this point in 2024 and 2022; the "votes still to come" settings slide between those two patterns.

How each group votes. For 2022 and 2024, a statistical model of every precinct's results estimates how each group voted, letting the rate vary with the precinct's primary mix and keeping each precinct's own lean. Most races use the average of the countywide judicial races, close to a pure party vote. County Judge uses the 2022 County Judge race; statewide offices use the same office in 2022 (and the 2024 Senate, presidential or Railroad Commission race). State House incumbents keep three-quarters of how far they ran ahead of or behind the party baseline in their last contested race.

Races that cross the county line. Five races here reach past Harris: CD 9 (all of Liberty County), CD 7 and CD 18 (parts of Fort Bend) and the 1st and 14th Courts of Appeals (nine nearby counties). Each outside county is built the same way from Texas Secretary of State lists: every early and mail voter in 2022, 2024 and 2026 is matched to the state's March primary lists, and the county's own 2022 and 2024 late votes give its two turnout patterns (the state lists have no Election Day voters, so those are filled in from county totals in proportion to Harris's Election Day patterns by group). Fort Bend is split by precinct on the 2025 map, using the Texas Legislative Council's precinct reports. Each county's groups vote as Harris's model says for a place with its primary mix, shifted by one amount so the model matches that county's (or Fort Bend part's) statewide judicial races in 2022 and 2024. Rebuilt on 2024's results, CD 9 comes out Trump +20, matching the Legislative Council's figure. A separate UH Hobby School survey of CD 9 is blended in the same way as the countywide poll.

Projected wins and uncertainty. A projected win is the candidate ahead in the central estimate. Chances come from simulated elections that each draw a countywide error shared by every race, an error for each race, and a turnout pattern near your settings. Fitting one year and predicting the other missed the county by about 2 points either way, so the shared error is set wider than that.

Open seats and flips. "(Open)" means the person holding the seat is not on the ballot. The holder is the last elected winner from the county's canvass, corrected for known appointments (Governor's appointees to district courts, Commissioners Court appointees, the County Attorney) and the current justices of the peace. A highlighted "Flip" is a seat where the projected winner's party differs from the holder's. Congressional holders are from the old lines; statewide offices show open seats only.

What it leaves out. Candidates' campaigns, except through incumbency and the polls. Third-party candidates (shares are two-party). Undervotes. Precinct-level detail stays off this page.

—