2028 Senate swingometer
Senate 2028
Share
51 for control
Loading
Democratic lead
Republican lead
Towards D
Towards R
Each race: its state's 2024 presidential margin and the holder's past edge, moved by the national vote.
Sources
Method: each 2028 race starts from its state's 2024 presidential two-party margin (the swingometer's calibrated state), plus k times the holder's candidate effect, moved by the national vote against the 2024 presidential vote (two-party R+1.5) and by any nudge or call you set. Control: 51 seats, or 50 with the vice president, who comes from the 2028 presidential race.
Seats not up in 2028: senators serving now (unitedstates/congress-legislators, CC0) and the 2026 Class II winners from AP, DDHQ and civicAPI calls, not certified results. Florida and Ohio: the 2026 special elections' called winners.
Senate after 2026 (the headline changes and the changes-party outlines): those seats not up plus the Class III holders, D 56 R 44; calls, not certified. Certified results will replace the calls when there are some.
Race polls: none of the 2028 Senate races has been polled yet; a race's card will list its polls when there are some.
Retirements: none announced as of 2026-11-05 (checked: news search for 2028 Senate retirements; English Wikipedia, 2028 United States Senate elections (revision 1383921898, 5 Nov 2026), which lists no retirements).
National vote default: the 2026 House vote as counted to 2026-11-19, summed from the district counts on the 2026 results map over the 411 districts with a Democrat and a Republican on the ballot (72 complete; official state feeds, then civicAPI; count so far, not certified), D 55.3 R 43.3, two-party D+12.2. Left out: districts without both a Democrat and a Republican (same-party and uncontested races), districts with no count, and Louisiana (an all-party November ballot). A 2026 result, not a 2028 forecast. Pills: the 2026 generic ballot average and the newest polls in it, adapted from Wikipedia, revision 1383604663 (CC BY-SA 4.0); each loads its two-party margin into the national slider only.
Ranges: 2,000 seeded simulations at these settings, one shared national error (SD 3.33) and an error per seat (SD 5.8599): the 2026 midterms model's Senate uniform-swing error (us-midterms-2026/data/senate.json meta.uniform_swing_sigma.senate; national part min(senate_odds.json national_error_sd, sigma) as build_senate_odds.py). They are 2026 errors applied two years early, so the real uncertainty is wider. A seat set by hand takes no error.
Candidate effect: an incumbent's last result minus the presidential margin in the seat, less that year's chamber average, carried over at k = 0.25 (backtest bounds 2018-2024: 0.25-0.5). Senate: MEDSL 1976-2024 (CC0); none where the holder changed or there was no two-party result (AK).
2024 results by state: MIT Election Data and Science Lab (CC0 1.0). State boundaries: us-atlas 3.0.1 (ISC licence).