Skip to content
Term: --%

The midterm reckoning / 3 November 2026

He lost Congress. The bill came due.

Trump asked voters to pretend he was on the ballot. They obliged. Democrats won control of both chambers, and the president who sold himself as the solution became the problem his candidates could not outrun.

Reviewed 16 November 2026. Results snapshot: 5 November 2026 US reporting.

The short version

The result. In 45 seconds.

Results first. Then our top 10. Our ranking, not a poll.

Silent video. Optional on-screen text captions.

2026 midterms results reel. Text and graphics only; no audio.

Video snapshot: . Reported race calls, not certified final results.

The full results, reasons and source notes are below. No need to pause on the small print.

The animated explainer

Ten reasons. Less restraint.

Rubber stamps and flying banknotes. The sources still have to stand up. Our ranking, not a poll.

Silent video. Optional on-screen text captions.

2026 midterms top 10 explainer. Text and graphics only; no audio.

Video snapshot: . Reported race calls, not certified final results.

The full results, reasons and source notes are below. No need to pause on the small print.

01 / The result

Both chambers. Not a rounding error.

By 5 November, seven Republican-held Senate seats had flipped: North Carolina, South Carolina, Ohio, Texas, Kansas, Iowa, Maine.

A dated snapshot of reported race calls, not certified final results or an automatically updating election feed. The Decision Desk HQ snapshot, checked 5 November 2026 at 8:49pm UTC, put the Democratic caucus at 54 seats (including two independents who caucus with Democrats), Republicans at 44, with Alaska and Mississippi uncalled. These figures are deliberately frozen to that 5 November snapshot; later race calls are not included.

Sources: NPR: Democrats win both chambers.

02 / Evidence before adjectives

Our ranking is not the poll.

This is our editorial ranking of 10 contributing issues, campaign failures and electoral patterns, not a statistical ranking of why each voter voted. Where themes overlapped, we merged them: petrol and power bills sit under the cost of living, the farm belt under tariffs. Iran, tariffs and ICE lead our list. The AP Voter Poll's top-issue question instead put the economy first (39%), democracy second (32%), immigration third (14%) and healthcare fourth (8%). Different questions measure different things. These reasons still overlap: do not add their percentages, costs or supposed effects together.

Economy39%
State of democracy32%
Immigration14%
Healthcare8%

Selected responses to one top-issue question. The four bars total 93%; other responses are not shown. The scale is 0 to 100%, not normalised to these four answers.

Sources: AP via PBS: voters' top issues; Philenews: AP top-issue percentages.

03 / The list

The top 10 reasons Trump lost Congress.

Ten distinct arguments, not twenty variations. Where reasons repeated a theme, we merged them and listed what each one folds in. Policy failures first, then accountability, then the candidates, coalition and campaign machinery that could not save him. Each entry tells you what the evidence can carry, and what it cannot.

  1. Policy / Voter survey + reported costs

    1. Iran: the peace president delivered another war.

    About 6 in 10 voters said military action against Iran was the wrong decision

    Folds in: War casualties and cost; The broken peace promise.

    The war was not an abstract foreign-policy argument. It meant casualties, expensive fuel and a promise of peace that kept requiring more bombs. AP reports roughly one-third of voters called the Iran war crucial to their congressional vote, and cited a Pentagon cost estimate above US$47.1 billion. Stars and Stripes reported 23 US military deaths and 865 wounded on 29 October, combining two Pentagon operational categories and including noncombat deaths. Even his own side stopped spinning it. JD Vance posted, "The reason we lost is the war." Trump defended his Iran policy while blaming petrol costs. During Fox's election-night coverage, Trump told Bret Baier on 3 November 2026, "People agree with what I'm doing on Iran, but they really probably weren't willing to do the short-term consequence of paying more for gasoline."

    The limit: Our number one, not the electorate's number-one answer to the top-issue question. The casualty figure is a dated operational total, not 23 combat deaths, and the war-cost estimate is not a live clock. Vance's and Trump's explanations are politicians' verdicts, not measurements.

    Sources: AP via PBS: oversight, Iran, ICE and healthcare; PBS transcript: Trump's inaugural peace pledge; Stars and Stripes: Pentagon casualty categories; Roll Call: AP flip count, uncalled races and the GOP blame game; Fox News: Trump's election-night explanation to Bret Baier.

  2. Policy / Voter survey + economic model + results

    2. Tariffs: the tax he kept pretending somebody else paid.

    About two-thirds said Trump's tariffs hurt the US economy

    Folds in: Household tariff costs; The farm-belt squeeze (Iowa and Kansas flipped).

    He promised protection and delivered a bill. AP found voters blamed his tariffs, alongside Iran and the broken promise of lower prices, for their economic pain. Yale's 9 November correction estimates annual household costs of about US$900 under current law, an ultimate consumer-price impact of about 0.6% and an 11.2% average statutory tariff rate (11.4% by year-end). Farmers got it from both ends: Brookings described lost Chinese crop sales on top of dear fuel, fertiliser and credit, and Republicans then lost Senate seats in Iowa and Kansas. Former Republican senator Pat Toomey put it plainly: "Arbitrary and capricious tariffs directly raised costs for all American consumers."

    The limit: Yale's model is a policy scenario, not an invoice paid by every household, and an average statutory import rate is not a product's retail-price increase. Brookings' piece was pre-election commentary; a statewide flip does not prove farmers caused it, or split switching from turnout.

    Sources: AP via PBS: voters' top issues; AP via PBS: how Republicans lost; AP transcript via PBS: Trump's tariff claim; Yale Budget Lab: November tariff model, corrected 9 November; Yale: corrected release data and key takeaways; Brookings commentary: rural economic pressure; NPR: Democrats win both chambers; Roll Call: AP flip count, uncalled races and the GOP blame game.

  3. Policy / Voter survey

    3. ICE: enforcement became overreach.

    57% said immigration enforcement had gone too far

    The raids and the footage did not look like competence to a majority of voters. Newsweek reports 60% disapproved of ICE, and 89% of voters who thought enforcement had gone too far backed Democrats. That is a substantial backlash. It also deserves its awkward footnote: Republicans still led 44% to 41% on which party voters trusted to handle immigration.

    The limit: Only 14% named immigration the country's top issue. Voters calling enforcement crucial actually favoured Republicans, 58% to 41%. Disliking ICE's methods does not prove immigration was the decisive national cause.

    Sources: Newsweek via Yahoo: ICE backlash and its limits; AP via PBS: oversight, Iran, ICE and healthcare.

  4. Policy / Voter survey + campaign reporting

    4. Cost of living: the checkout, the bowser and the power bill.

    About 8 in 10 said inflation had caused their household moderate or severe hardship

    Folds in: Grocery prices; Petrol costs; Data centres and electricity bills.

    He won office selling relief. By the midterms, voters were still waiting at the checkout. The economy topped AP's issue question at 39%, and roughly six in ten voters blamed Trump's policies more than outside factors for high prices. Then came the power bill. CNBC reported AdImpact counted US$173 million in general-election ads about data centres, almost all negative, and Democrats who campaigned against them in Ohio and South Carolina won.

    The limit: A voter's attribution of blame is not a causal estimate. Ad spending measures campaign emphasis, not persuasion; AP found data centres less likely to be crucial than corruption or prices. The explainer's petrol bowser is an illustration, not a quoted fuel price.

    Sources: AP via PBS: voters' top issues; AP via PBS: oversight, Iran, ICE and healthcare; Philenews: AP top-issue percentages; AP via PBS: how Republicans lost; CNBC: Musk spending and the data-centre backlash.

  5. Accountability / Voter survey + advocacy-commissioned pre-election poll

    5. Self-dealing stopped being background noise.

    More than half called government corruption crucial to their congressional vote

    Folds in: Corruption; Personal priorities over the country's; The Epstein files.

    AP found corruption was crucial to slightly more voters than the cost of living. About six in ten said Trump put his personal interests ahead of the country's, and AP's analysis ties that view to family business deals, cryptocurrency and the White House ballroom. You can only talk about your own grandeur for so long to people worried about an electricity bill. The Epstein files gave the argument a recognisable test: would powerful people face scrutiny, or another delay?

    The limit: These are voters' perceptions and reported activity, not criminal findings against any person or transaction. The Epstein evidence is an advocacy-commissioned GBAO pre-election poll, not an AP election-day finding, and being mentioned in a file is not proof of criminal conduct.

    Sources: AP via PBS: how Republicans lost; AP via PBS: voters' top issues; COURIER/Reckoning Action: commissioned GBAO pre-election poll.

  6. Accountability / Voter survey

    6. An unpopular president with a rubber-stamp Congress.

    About 70% wanted Congress to provide more checks on the president

    Folds in: Checks and balances; Trump's own unpopularity.

    About six in ten voters disapproved of how Trump was handling the job, and more than half doubted he was a strong leader or fought for people like them. Democracy ranked second on the top-issue question. Around seven in ten said Congress was not doing enough to check the president, including roughly four in ten Republicans. Handing Democrats both chambers made sense to voters who wanted restraint rather than another round of applause.

    The limit: A demand for oversight is not automatic support for impeachment or every Democratic policy. These are voter assessments, not medical diagnoses, and election-survey approval is a different measure from the site's rolling polling average.

    Sources: AP via PBS: oversight, Iran, ICE and healthcare; AP via PBS: how Republicans lost; AP via PBS: voters' top issues; Philenews: AP top-issue percentages.

  7. Policy / Voter survey + policy reporting

    7. Healthcare got dearer, and the 'beautiful' bill had ugly fine print.

    About half said Trump's policies worsened healthcare access and costs

    Folds in: Premiums after the ACA credits lapsed; Medicaid cuts.

    Healthcare did not need to win the top-issue question to decide whether a household trusted the government. The enhanced ACA premium tax credits expired at the end of 2025. Separately, the signature tax law carried a projected reduction of nearly US$1 trillion in federal Medicaid spending over a decade, and the Christian Science Monitor reported candidates campaigning on cuts and rural-clinic fears. A tax cut can be real and still lose the argument when people fear losing their cover.

    The limit: Premium changes vary by plan, income and state. The Medicaid figure is a ten-year projection, not money already removed in 2026, and several provisions, including work requirements, start after the midterms.

    Sources: AP via PBS: oversight, Iran, ICE and healthcare; Christian Science Monitor: tax law and Medicaid campaign.

  8. Campaign / State voter surveys + campaign reporting

    8. Trump-first candidates, voter-last results.

    About two-thirds of Texas Senate voters said Ken Paxton lacked the character to serve

    Folds in: Weak nominees; Loyalty to Trump as a liability.

    AP's state findings are unusually blunt. About six in ten South Carolina Senate voters said Darline Graham lacked the experience. About six in ten in North Carolina said Michael Whatley was too supportive of Trump; Roy Cooper beat him. Ken Paxton, Darline Graham and Michael Whatley all lost their Senate races. Trump told supporters to pretend he was on the ballot, then said he was not on it after the losses. An interesting late adjustment to the methodology.

    The limit: State findings cannot be turned into a national percentage, and the surveys show association, not the isolated effect of an endorsement. Paxton was impeached by the Texas House and acquitted by the Senate, not convicted.

    Sources: AP via PBS: how Republicans lost; NPR: Talarico defeats Paxton in Texas; NPR: Andrews defeats Graham in South Carolina; WUNC via NPR: Cooper defeats Whatley in North Carolina; The Independent: Trump's before-and-after ballot claim.

  9. Electorate / Voter surveys + electoral pattern

    9. The 2024 coalition was a lease, not a freehold.

    Roughly 6 in 10 Hispanic voters backed Democratic House candidates

    Folds in: Latino voters; Young men and non-college youth.

    Newsweek's account of AP's survey has roughly six in ten Hispanic voters backing Democrats and about one-third Republicans. Texas Republicans drew their new map assuming Trump's 2024 Latino gains would hold. They did not. Tufts CIRCLE reports 59% of young men, 70% of young women and 60% of youth without college experience backed Democrats, with a Democratic edge among white youth too. Treating a diverse electorate as the party's new property was the mistake.

    The limit: A midterm electorate is not a presidential one, and these figures do not prove the same people switched. CIRCLE uses NORC's separate survey, its early estimates may shift, and the shares are among voters, not all young adults.

    Sources: Newsweek via Yahoo: ICE backlash and its limits; Texas Tribune: the redraw backfires; Tufts CIRCLE: youth vote by demographic group.

  10. Campaign / Reported results + campaign-finance reporting

    10. Neither the maps nor Musk's millions could save them.

    21 of 52 Democratic wins in Trump-won House seats came in states that redrew their maps

    Folds in: The mid-decade gerrymander; Musk's spending.

    Republicans redrew maps to manufacture seats. Roll Call's AP count found 21 of the 52 Democratic wins in Trump-won districts came in states that had redrawn, including six in Texas. The Texas Tribune reports the expected 30-8 Republican delegation became 24-14. Elon Musk spent at least US$474 million on the 2026 midterms, US$224 million of it in Texas, where Paxton and Abbott lost. By Wednesday morning, CNBC counted 29 losses and three wins among the 44 federal races his America PAC funded.

    The limit: This explains why a protective strategy failed, not a separate grievance. District counts are race-call snapshots, not certified returns. The PAC figures are a subset of Musk's total, and losing a heavily funded race does not prove the money caused the loss.

    Sources: Roll Call: AP flip count, uncalled races and the GOP blame game; Texas Tribune: the redraw backfires; CNBC: Musk spending and the data-centre backlash.

04 / What changes

Less blank cheque. More checks.

New Congress, not a new president.

The incoming Congress takes office in January 2027. Democrats gain control of committees, oversight and the legislative agenda. The election does not remove Trump from office or instantly reverse his executive actions.

Subpoenas and confirmations matter.

Democratic committee chairs can investigate the administration. Senate control also gives Democrats leverage over judicial and executive nominations. Oversight is a power, not a guarantee that every demand succeeds.

The veto is still there.

Trump retains the presidential veto. Ordinary legislation, spending deals and efforts to reverse policy still face constitutional and Senate procedural hurdles. Winning Congress is not the same thing as passing everything.

A rejection is not a blank cheque.

AP found only about half of voters viewed the Democratic Party favourably, and around two in ten voters who supported Democratic candidates disliked the party. Democrats won a demand for change. Now they have to do something with it.

Sources: NPR: Democrats win both chambers; AP via PBS: how Republicans lost.

He said, “I wasn't on the ballot.” True, literally. Less convincing after asking voters to pretend he was.

Sources: The Independent: Trump's before-and-after ballot claim.

05 / The receipts

Show the working.

The AP Voter Poll was conducted by SSRS using probability-based and non-probability-based interviews with registered voters in all 50 states, ending 3 November. AP reports a full-sample margin of sampling error of +/-0.6 percentage points; subgroup uncertainty is larger. CIRCLE uses the separate NORC Civic Health & Elections Project. Early survey estimates may be revised. Pre-election polls and commentary are labelled separately.

All money on this page is in US dollars. Counts, price observations and model estimates retain their own dates. The locked Year 1 scorecard is not silently rewritten as a November scorecard.

We have an opinion. We also have an obligation not to inflate the numbers to make it louder. The data-health page shows overdue editorial reviews elsewhere on the site.

  1. NPR: Democrats win both chambers (4 November 2026)
  2. NPR: Talarico defeats Paxton in Texas (3 November 2026)
  3. NPR: Andrews defeats Graham in South Carolina (3 November 2026)
  4. WUNC via NPR: Cooper defeats Whatley in North Carolina (3 November 2026)
  5. AP via PBS: voters' top issues (3 November 2026)
  6. AP via PBS: oversight, Iran, ICE and healthcare (3 November 2026)
  7. AP via PBS: how Republicans lost (4 November 2026)
  8. Philenews: AP top-issue percentages (4 November 2026)
  9. Newsweek via Yahoo: ICE backlash and its limits (4 November 2026)
  10. Yale Budget Lab: November tariff model, corrected 9 November (9 November 2026)
  11. Yale: corrected release data and key takeaways (9 November 2026)
  12. Stars and Stripes: Pentagon casualty categories (29 October 2026)
  13. Christian Science Monitor: tax law and Medicaid campaign (18 August 2026)
  14. COURIER/Reckoning Action: commissioned GBAO pre-election poll (3 September 2026)
  15. Tufts CIRCLE: youth vote by demographic group (4 November 2026)
  16. Brookings commentary: rural economic pressure (19 October 2026)
  17. Texas Tribune: the redraw backfires (4 November 2026)
  18. CNBC: Musk spending and the data-centre backlash (4 November 2026)
  19. The Independent: Trump's before-and-after ballot claim (4 November 2026)
  20. Roll Call: AP flip count, uncalled races and the GOP blame game (5 November 2026)
  21. Fox News: Trump's election-night explanation to Bret Baier (4 November 2026)
  22. Path to Win: Decision Desk HQ Senate results snapshot, checked 5 November 2026 at 8:49pm UTC (5 November 2026)
  23. Christian Science Monitor: five flips and projected 2027 governorships (5 November 2026)
  24. PBS transcript: Trump's inaugural peace pledge (20 January 2025)
  25. AP transcript via PBS: Trump's tariff claim (24 February 2026)

Read the report data (JSON) · Scoring methodology · Current statistics