Sign inBook a demo

2026 Midterms

Sealed October 28, 2026 · v2.1

Who wins Congress, race by race.

Democrats have a 95% chance of winning the House and 68% of the Senate. We simulated voters in every state, combined their answers with the polls, and ran the election 20,000 times. Polls collected through October 27, 2026.

Control of Congress

20,000 simulated elections

Both chances rose since the October 23 version: the Senate from 49.8% as published on October 23 (44.2% after the October 24 Alaska poll correction), the House from 54.6%. See what changed and why

Democrats have a 68% chance of controlling the Senate.

Chance of control, from 20,000 simulated elections: Democrats 68%, Republicans 28%, neither party alone 4% (an independent decides). The seat counts below are median results, a separate measure from the chance of control.

DEM 68% chanceREP 28% chance
Median seats52 DEM51 seats to win ▾Median seatsREP 48
45 D + 2 I seats today53 seats today

Democrats need 51 seats. Republicans need 50, because the Vice President breaks ties.

35 of 35 races forecast · 65 seats not on the ballot · median result shown at each end

  • Safe D 10
  • Likely D 3
  • Lean D 4
  • Toss-up 3
  • Lean R 2
  • Likely R 3
  • Safe R 10

Democratic seats across 20,000 simulated elections

Median 52 · most likely 52 · 80% of results fall between 49 and 54

44 seatsBlue: Democrats hold a majority57 seats

The map

35 races · 15 states with no Senate race

The Senate map

Alaska: Republican favored, 50%, Toss-upAlabama: Republican favored, >99%, Safe RArkansas: Republican favored, 98%, Safe RArizona: no Senate race this yearCalifornia: no Senate race this yearColorado: Democrat favored, 99%, Safe DConnecticut: no Senate race this yearDelaware: Democrat favored, >99%, Safe DFlorida: Republican favored, 84%, Likely RGeorgia: Democrat favored, 94%, Likely DHawaii: no Senate race this yearIowa: Republican favored, 53%, Toss-upIdaho: Republican favored, >99%, Safe RIllinois: Democrat favored, 99%, Safe DIndiana: no Senate race this yearKansas: Republican favored, 57%, Toss-upKentucky: Republican favored, 98%, Safe RLouisiana: Republican favored, 92%, Likely RMassachusetts: Democrat favored, >99%, Safe DMaryland: no Senate race this yearMaine: Democrat favored, 64%, Lean DMichigan: Democrat favored, 77%, Lean DMinnesota: Democrat favored, 88%, Likely DMissouri: no Senate race this yearMississippi: Republican favored, 97%, Safe RMontana: Republican favored, 91%, Likely RNorth Carolina: Democrat favored, 96%, Safe DNorth Dakota: no Senate race this yearNebraska: Republican favored, 67%, Lean RNew Hampshire: Democrat favored, 90%, Likely DNew Jersey: Democrat favored, 98%, Safe DNew Mexico: Democrat favored, >99%, Safe DNevada: no Senate race this yearNew York: no Senate race this yearOhio: Democrat favored, 73%, Lean DOklahoma: Republican favored, >99%, Safe ROregon: Democrat favored, 99%, Safe DPennsylvania: no Senate race this yearRhode Island: Democrat favored, >99%, Safe DSouth Carolina: Republican favored, 67%, Lean RSouth Dakota: Republican favored, >99%, Safe RTennessee: Republican favored, >99%, Safe RTexas: Democrat favored, 65%, Lean DUtah: no Senate race this yearVirginia: Democrat favored, >99%, Safe DVermont: no Senate race this yearWashington: no Senate race this yearWisconsin: no Senate race this yearWest Virginia: Republican favored, >99%, Safe RWyoming: Republican favored, >99%, Safe R
  • Safe D
  • Likely D
  • Lean D (striped)
  • Toss-up
  • Lean R (striped)
  • Likely R
  • Safe R
  • No race this year (dotted)
Timeline

The timeline changes the overview and this map. Everything below the map is the final version, v2.1.

Ratings

Safe, likely, lean, toss-up

Senate ratings

Lean Democratic

Total 4

Toss-up

Total 3 · DEM 0 · REP 3

Lean Republican

Total 2

Key races

8 closest statewide races

Key races, and which voter groups back each candidate.

How each simulated voter group would vote, and the issues it ranks highest.

Alaska Senate: Dan S. Sullivan vs. Mary Peltola

Dan S. Sullivan wins in 50% of simulated elections · first-choice vote share 45.3% to 48.7%. Ranked choice: if no one wins a majority of first choices, later choices are counted, so the first-choice leader can lose.

Dan S. Sullivan (R)Mary Peltola (D)
All voters

Party

Democrats
Independents
Republicans

Gender

Men
Women

Age

18–29
30–44
45–64
65+

These are our simulated voter groups alone, before the polls are combined in, so they can lean differently from the forecast.

Top issues for simulated Alaska voters

Cost of living and the economy35%
Immigration and the border13%
Democracy and the rule of law13%
Taxes and government spending9%

Strongest groups for Sullivan

Men54–36
65+55–38
No degree53–38

Strongest groups for Peltola

Black77–17
Other race60–31
College grads55–39
Ask a simulated audience your own question

Closest races

Projected margin under each bar

Closest Senate races

  • AlaskaD Mary PeltolaR Dan S. Sullivan50%First-choice margin D +3.450%Sullivan ▼15
  • IowaD Josh TurekR Ashley Hinson47%Projected margin R +0.453%Hinson ▼8
  • KansasD Adam HamiltonR Roger Marshall43%Projected margin R +1.257%Marshall ▼4
  • MaineD Troy D. JacksonR Susan M. Collins64%First-choice margin D +2.236%Jackson ▲12
  • TexasD James TalaricoR Ken Paxton65%Projected margin D +2.235%Talarico ▲14
  • NebraskaR Pete RickettsI Dan Osborn67%Projected margin R +2.733%—
  • South CarolinaD Annie AndrewsR Darline Graham33%Projected margin R +2.967%Graham ▼11
  • OhioD Sherrod BrownR Jon Husted73%Projected margin D +3.927%Brown ▲13
  • MichiganD Abdul El-SayedR Mike Rogers77%Projected margin D +4.623%El-Sayed ▲16
  • FloridaD Angie NixonR Ashley Moody16%Projected margin R +6.384%—

National outcomes

House and Senate counted in the same 20,000 elections

Democrats win both chambers in about 2 in 3 simulated elections.

  • Democrats sweep67.6%
    13,513 of 20,000 elections
  • Democratic House, Republican Senate23.6%
    4,711 of 20,000 elections
  • Republicans sweep4.1%
    812 of 20,000 elections
  • Republican House, Democratic Senate0.4%
    72 of 20,000 elections
  • No clear majority in one chamber4.5%
    892 of 20,000 elections

All 20,000 simulated elections, by who controls what

Each square is 50 simulated elections.

Who wins the House

Democratic Party95.1%
Republican Party4.5%

Neither party has 218 in 82 elections

Who wins the Senate

Democratic Party68.0%
Republican Party27.9%

Independents decide the other 811

How many Senate seats will Democrats hold?

48 or fewer9.5%
49–5022.5%
51–5235.3%
53 or more32.7%

Including the 34 seats not up

How many House seats will Democrats win?

Under 2184.9%
218–23422.2%
235–25447.9%
255 or more25.0%

Of 435 seats

What if the polls are off?

The same 20,000 elections, rerun under each assumption. Only the first line is the forecast. If every poll overstates Democrats by 3 points, Democrats still win the House in 91% of those elections. The last line is simulated voters alone, with no 2026 polls, shown for comparison only.

AssumptionSenate (D)House (D)
The forecast: simulated voters combined with the polls68%95%
Polls overstate Democrats by 2 points51%93%
Polls overstate Democrats by 3 points42%91%
Simulated voters alone (no 2026 polls), for comparison, never part of the forecast28%76%

Since October 23

v2.0 to v2.1

Rating changes since the last version.

21 favorites changed and 241 ratings moved between the October 23 forecast (v2.0, with the October 24 Alaska poll correction) and the final one (v2.1), across all 646 contests.

Correcting a Republican tilt

The simulation favored Republicans more than the polls did: by 8.4 points in Senate races, 7.1 in House races and 5.9 in governor races. We now correct for that gap. Races without polls moved most.

Narrower range of outcomes

The forecast's uncertainty is now calibrated on races from 1998 to 2024 and tested on 506 races from 2018 to 2024. Democrats win 49 to 54 Senate seats in 8 of 10 simulations, against 44 to 55 before, so a small lead now gives a higher chance of winning.

Newer polls

Polls collected through October 27. Our first forecast, of October 15, had the same favorite as this one in 49 of the 56 races both cover.

Likely D → Safe D

D Colorado Gov97% ▲7
D Colorado Sen99% ▲13
D Delaware Sen>99% ▲7
D North Carolina Sen96% ▲9
D New Jersey Sen98% ▲14
D New York Gov97% ▲8
D Oregon Sen99% ▲9

Lean D → Likely D

D Georgia Sen94% ▲14
D Michigan Gov90% ▲12
D Minnesota Sen88% ▲20
D New Hampshire Sen90% ▲14
D New Mexico Gov90% ▲11

Lean R → Toss-up

R Alaska Sen50% ▼15
R Georgia Gov53% ▼8
R Iowa Sen53% ▼8
R Kansas Sen57% ▼4

Toss-up → Lean D

D Maine Sen64% ▲12
D Texas Sen65% ▲14
D Wisconsin Gov71% ▲19

Likely R → Safe R

R Kentucky Sen98% ▲4
R Mississippi Sen97% ▲9

Lean R → Likely D

D Iowa Gov84% ▲47

Lean R → Lean D

D Ohio Gov64% ▲44

Where the favorite changed

  • Arizona District 2Now Jonathan Nez, 53% · Toss-upWas Eli Crane at 56% · Toss-up
  • California District 34Now Angela Gonzales-Torres, 54% · Toss-upWas Jimmy Gomez at 89% · Likely D
  • California: Proposition 45Now No, 53% · Toss-upWas Yes at 65% · Lean Yes
  • Colorado District 8Now Manny Rutinel, 66% · Lean DWas Gabe Evans at 57% · Toss-up
  • Colorado: Proposition 135Now No, 55% · Toss-upWas Yes at 51% · Toss-up
  • Iowa GovernorNow Rob Sand, 84% · Likely DWas Zach Lahn at 64% · Lean R
  • Iowa District 2Now Lindsay James, 53% · Toss-upWas Joe Mitchell at 55% · Toss-up
  • Iowa District 3Now Sarah Trone Garriott, 73% · Lean DWas Zach Nunn at 58% · Toss-up
  • Idaho: HJR 4Now No, 52% · Toss-upWas Yes at 50% · Toss-up
  • Kentucky District 6Now Ralph A. Alvarado, 75% · Lean RWas Zach Dembo at 52% · Toss-up
  • Massachusetts: Question 3Now No, 68% · Lean NoWas Yes at 75% · Lean Yes
  • Michigan: Proposal 26-1Now No, 84% · Likely NoWas Yes at 88% · Likely Yes
  • Missouri District 2Now Fred Wellman, 56% · Toss-upWas Ann Wagner at 53% · Toss-up
  • Missouri: Proposition ANow No, 55% · Toss-upWas Yes at 68% · Lean Yes
  • Nevada GovernorNow Aaron D. Ford, 56% · Toss-upWas Joseph Lombardo at 58% · Toss-up
  • New York District 1Now Christopher J. Gallant, 53% · Toss-upWas Nicholas J. LaLota at 78% · Lean R
  • New York District 17Now Caitlin Conley, 61% · Lean DWas Michael Lawler at 51% · Toss-up
  • Ohio GovernorNow Amy Acton, 64% · Lean DWas Vivek Ramaswamy at 79% · Lean R
  • Ohio District 7Now Max Miller, 59% · Toss-upWas Brian Poindexter at 74% · Lean D
  • Texas District 32Now Jace Yarbrough, 65% · Lean RWas Dan F. Barrios Jr. at 50% · Toss-up
  • Virginia District 1Now Shannon Leigh Taylor, 52% · Toss-upWas Robert J. "Rob" Wittman at 54% · Toss-up

All races

Closest first · 35 shown

Every Senate race

State by state

50 states

State by state

Every run

October 15 to the final version

Every forecast we made, and the other ways we ran it.

Each version was kept as it was made, with other methods run beside it, so all of them can be scored against the results: the October 15 forecast was sealed that day, and v1.0.0 was frozen with a checksum. The bold line of each version was its forecast. The first two versions covered 56 contests in 17 states (candidate races in 8 states, ballot measures in 15) and did not forecast control of Congress.

VersionRunRacesSame favorite as v2.1Senate (D)House (D)
Oct 15Simulation + pollsThe published forecast: simulated voters given past results and the 2026 polls, then combined with the polling average.5649 of 56——
Simulation, no pollsSimulated voters given past results only, no 2026 polls.5639 of 56——
Simulation + research briefSimulated voters given past results and a research brief, added once in the week the forecast was sealed.5640 of 56——
Simulation + local community commentsSimulated voters also given comments from each state's local online communities, sealed the same day.5649 of 56——
Gemini aloneThe underlying language model, Gemini, prompted directly, without Planetarium's simulated voter groups.5641 of 56——
Polling averagePolling averages alone: published averages where there are any, our own weighted average elsewhere.3830 of 38——
Past resultsThe 2024 result of each state or district; measures 50-50.5626 of 56——
v1.0.0Forecast (B where a state had community records, otherwise A)Vote shares only; no win chances were part of this version.4634 of 46——
B: facts, past results and community recordsOfficial facts and past results plus up to 12 reviewed community records per state.3122 of 31——
A: facts and past resultsB without the community records.4634 of 46——
R: standard survey simulationPlanetarium's standard survey simulation, run as a reference for vote shares.4634 of 46——
v2.0The forecast: simulated voters combined with the pollsThe published forecast.646625 of 64644.2%54.6%
Simulated voter groups + pollsOne of the two simulated estimates, combined with the polls.646623 of 64643.8%52.9%
Simulated turnout model + pollsThe other simulated estimate, combined with the polls.643610 of 64344.4%56.0%
Simulated voter groups aloneNo polls.646598 of 64612.2%33.5%
Simulated turnout model aloneNo polls.643588 of 64314.1%37.7%
Polls aloneOur own polling average, where a race has polls.202180 of 20254.9%63.3%
v2.1The forecast: simulated voters combined with the pollsThe published forecast.646646 of 64668.0%95.1%
Simulated voter groups + pollsOne of the two simulated estimates, combined with the polls.646638 of 64668.4%95.0%
Simulated turnout model + pollsThe other simulated estimate, combined with the polls.642623 of 64267.6%95.3%
Simulated voter groups aloneNo polls.646597 of 6466.3%26.8%
Simulated turnout model aloneNo polls.642587 of 6427.0%27.5%
Polls aloneOur own polling average, where a race has polls.214192 of 21462.5%73.6%
Planetarium aloneSimulated voters given the 2025–26 results and no polls; never part of the forecast.646615 of 64628.4%75.9%

Senate and House: the chance Democrats win control, in 20,000 simulated elections. v2.0 is shown with the Alaska poll correction; as published on October 23 its Senate figure was 49.8%. Where a run had no estimate for a seat, the simulation used that seat's past result.

How it works

Sealed before election day

23

simulated voter groups

In the 45 states with a Senate or governor race, split by party, age, gender, area, education and race. Each group got the same ballot question, naming every candidate and their party. House seats, and the 14 measures in the 5 states with no Senate or governor race, use one simulated group of all voters.

214

of 646 contests have polls

Where a race has polls, we combine our polling average with the simulation's estimate, once. In races without polls we still correct the simulation's Republican tilt, and the 321 House seats without polls shift together, by 2.69 points, toward the national generic-ballot polls.

20,000

simulated elections

Each with each state's own counting rules: ranked choice in Alaska and Maine, runoffs, and supermajority thresholds for measures.

v2.1

sealed and timestamped

Sealed on October 28, 2026, then scored against the results after November 3, race by race. See our accuracy record

Photo credits

Candidate photos from Wikimedia Commons, cropped and toned; those not listed are public domain, CC0 or released for free use (Iowa General Assembly portraits). Abdul El-Sayed: Conlan Houston, CC BY-SA 4.0; Adam Hamilton: Working Preacher, CC BY 4.0; Amish Shah: Gage Skidmore, CC BY-SA 3.0; Andy Biggs: Gage Skidmore, CC BY-SA 4.0; Ann Wagner: Official House photo, CC BY-SA 4.0; Brad Little: Phil White/Office of the Governor of Idaho, CC BY-SA 4.0; Brandon Herrera: Gage Skidmore, CC BY-SA 4.0; Bruce Blakeman: Arthur Raslich (Nassau County Photo Archives), CC BY-SA 4.0; Casey Askar: Gayonb, CC BY-SA 4.0; Charles Booker: Breadwin, CC BY-SA 4.0; Christine Drazan: Oregon State University, CC BY-SA 2.0; Cindy Holscher: Catboy69, CC BY 4.0; Dan Cox: MDGovpics (Maryland Governor's Office, Flickr), CC BY 2.0; Dan Osborn: Ramseywill, CC BY 4.0; Darren Bailey: Cobiblair, CC BY-SA 4.0; David Brock Smith: Oregon Secretary of State, CC BY 2.0; David Crowley: LaShawnda Jones, CC BY-SA 4.0; Dwayne Romero: Jeffrey Beall, CC BY 4.0; Fredrick Love: Arkansas Farm Bureau, CC BY 3.0; Gary Cordery: ThinkTech Hawaii, CC BY 3.0; Hallie Shoffner: Ark. Agricultural Experiment Station, CC BY-SA 4.0; Hannah M. Pingree: Hannah Pingree for Governor, CC BY 4.0; Helena Foulkes: Digitas Photos, CC BY 2.0; JB Pritzker: Oscar Garza, CC BY-SA 4.0; Jay Feely: Gage Skidmore, CC BY-SA 3.0; Jerri Green: Zach Kinslow, CC BY 4.0; Jim Pillen: Maryland GovPics (Polly Irungu), CC BY 4.0; Jocelyn Benson: JFM Photo 73, CC BY 4.0; John Braun: Troy Kirby, CC BY 4.0; John Deaton: Jeff Dranetz, CC BY 3.0; John James: United States Congress, CC BY-SA 4.0; Jonathan Nez: Wikiuser159597, CC BY-SA 4.0; Jonathan S. JKT Kreiss-Tomkins: James Poulson, CC BY-SA 4.0; Josh Shapiro: Maryland GovPics (Joe Andrucyk et al.), CC BY 2.0; Kathy Hochul: Metropolitan Transportation Authority (Don Pollard/Office of Governor Kathy Hochul), CC BY 4.0; Katie Hobbs: Gage Skidmore, CC BY-SA 4.0; Larry Rhoden: U.S. Embassy Jerusalem, CC BY 2.0; Manny Rutinel: Andybosselman, CC BY-SA 4.0; Michael Charles Echols: Timothy Holdiness, CC BY 4.0; Michael Whatley: Gage Skidmore, CC BY-SA 4.0; Michele Tafoya: Gage Skidmore, CC BY-SA 2.0; Ned Lamont: The Office of Governor Ned Lamont, CC BY-SA 4.0; Paige Gebhardt Cognetti: US Department of Labor, CC BY 2.0; Paul R. LePage: Gage Skidmore, CC BY-SA 3.0; Peggy Flanagan: Travellers & Tinkers, CC BY-SA 4.0; Ralph A. Alvarado: Gage Skidmore, CC BY-SA 3.0; Randy Villegas: Maremoto Fund, CC BY 4.0; Rebecca Cooke: UpNorthNews, CC BY 4.0; Rick Jackson: Cherie Martin, CC BY 4.0; Robert B. Charles: Hoveros, CC BY-SA 4.0; Robert J. "Bob" Harvie Jr.: Governor Tom Wolf (Flickr), CC BY 2.0; Roy Cooper: NetwonsBucket, CC BY-SA 4.0; Ryan Fazio: James Naleski, CC BY-SA 4.0; Sarah Huckabee Sanders: Gage Skidmore, CC BY-SA 4.0; Steve Hilton: Gage Skidmore, CC BY 2.0; Stewart Cathey Jr.: Timothy Holdiness, CC BY 4.0; Todd Achilles: DL a.k.a. LibertyDad, CC BY 4.0; Ty Masterson: NicholasGian25, CC BY-SA 4.0; Victor Marx: Gage Skidmore, CC BY-SA 2.0; Wes Moore: Maryland State Government, CC BY-SA 4.0; Amy Acton: Myselfsour2724, CC BY-SA 4.0; James Talarico: Larry D. Moore, CC BY 4.0; Ken Paxton: Gage Skidmore, CC BY-SA 4.0; Joseph Lombardo: Gage Skidmore, CC BY-SA 3.0; Vivek Ramaswamy: Gage Skidmore, CC BY-SA 4.0.

Simulate your own audience

Simulated audiences

Ask your own audience before you launch.

Start freeSee our accuracy record