This page tracks the 2026 US midterms as a running scoreboard: what is actually up for election, what the maps look like after an unprecedented redrawing season, and what the polling does and does not tell you. We update it as things change — the "Updated" date at the top tells you how fresh it is. * **Where things stand (August 19, 2026).** Florida voted on August 18, and the headline result was the least surprising thing on the ballot: **Byron Donalds** took the Republican nomination for governor with **47.8%** of the vote, with 99% of the expected vote counted. He was endorsed by Trump early and never trailed. Lt. Gov. Jay Collins had the sitting governor's machine behind him and it did not matter. That is the cleanest available measurement of what a Trump endorsement is currently worth in a Republican primary in the country's third-largest state, and the answer is: nearly everything (Washington Post, NBC News, August 18-19, 2026). * The genuinely interesting result was further down the ballot, and it cut the other way. In the Democratic primary for Florida's Senate special election, state Rep. **Angie Nixon** beat **Alexander Vindman** — the retired Army officer whose testimony helped drive Trump's first impeachment, and who had raised roughly **16 times** what she did. Money lost by a mile. Nixon now faces Republican Sen. **Ashley Moody** in November. Two incumbency stories rounded it out: Rep. **Cory Mills** lost his primary to Ryan Elijah, while Rep. **Randy Fine** held off a crowded field (Al Jazeera, August 19, 2026). * Read those two results together and the pattern is not "the establishment is losing." It is that endorsement and attention are doing the work money used to do, in both parties. Donalds converted a presidential endorsement into a plurality against better-funded institutional support; Nixon converted local standing into a win against a nationally famous, nationally funded opponent. Fundraising totals remain the number political reporting leads with, and this was a night where they explained very little. * On the polling, Decision Desk HQ's average on **August 14, 2026** put Democrats ahead **47-41**, a 6-point lead, back up from about 4 points in July. Their model treats a 6-point generic-ballot edge as making a House flip "much more likely." Hold that against the map section below before you get excited in either direction: the generic ballot measures votes, and this cycle's redistricting means votes and seats have come further apart than usual. **Where things stand (August 12, 2026).** Six states voted on August 11 — Alabama, Connecticut, Minnesota, South Carolina, Vermont and Wisconsin — and the result worth your attention is the one that did not finish. In South Carolina's special Republican Senate primary, **Sen. Darline Graham** finished first with **109,582 votes, 32.7%**. Graham was appointed by Governor Henry McMaster to the seat her brother Lindsey Graham held until he died in July 2026, weeks after winning the June nomination. Second place was a genuine fight: **Rep. Ralph Norman** on **82,523 (24.6%)**, **Rep. Russell Fry** on **65,732 (19.6%)** — close enough that several outlets called Graham's advance hours before they would call Norman's (NBC News, CBS News, August 11–12, 2026). Nobody reached 50%, so it goes to a **runoff on August 25**, and the winner meets Democratic nominee **Dr. Annie Andrews** on November 3. * Two things are worth separating here, because coverage tends to fuse them. A third of the vote for an appointed senator carrying the most famous surname in South Carolina Republican politics is a strong first-round showing — and it is also two-thirds of Republican primary voters choosing somebody else. Runoffs are where consolidation happens, and Norman starts the fortnight with Fry's 19.6% sitting unclaimed on the table. * Elsewhere on the night: Democrat **Rebecca Cooke** won Wisconsin's 3rd District primary **61–39** over Emily Berge; former Rep. **Jerry Carl** took Alabama's 1st District GOP special primary with **75%**; **Shomari Figures** won the Democratic nomination in Alabama's 2nd and will face Republican **Rhett Marques**; and Connecticut's Democratic incumbents — Courtney, DeLauro, Himes and Hayes — were all renominated, with **Amy Fogelstrom** winning the 1st District primary (Inside Elections, August 12, 2026). None of that moves the House math on its own. All of it fills in the names that the generic-ballot number is an abstraction of. * **Where things stand (August 6, 2026).** The generic ballot has barely twitched in months. As of August 5, 2026 the FiftyPlusOne aggregate puts Democrats on 49.6% and Republicans on 43.5% — a 6.1-point lead, sitting inside the same D+4 to D+7 band every other average has shown since spring. Morning Consult, which tracks it weekly, has had the Democratic lead between 3 and 4 points every week since the spring. When four averages of the same race disagree by three points and none of them move, the honest read is not "Democrats surge" or "Republicans close the gap" — it is that the national environment has been stable for a season and the coverage has to invent motion to have something to say. * One number under the topline is worth more than the topline. In the YouGov/Economist poll fielded August 1-4, 2026, independents split 38.2% Democratic to 26% Republican. Independents are the group that actually swings House seats, and a 12-point gap there is a wider margin than any of the headline averages — with the caveat that independents are also the group most likely to stay home, which is exactly what a generic-ballot question cannot measure. * **The map story nobody covers: Virginia.** The redistricting arms race is usually told as Texas versus California. Virginia is the case where the counter-punch failed, and it failed on a technicality. Voters approved an amendment in April 2026 to temporarily suspend the state's bipartisan redistricting commission and let the legislature redraw congressional maps mid-decade — Democrats answering Texas in kind. On May 8, 2026, the Supreme Court of Virginia struck the amendment down 4-3, ruling that the General Assembly had broken the state constitution's own procedure: its October 31, 2025 vote to advance the amendment came after early voting in the intervening House of Delegates election had already started, which Article XII does not permit. The US Supreme Court then declined an emergency request to revive it (Virginia Mercury, Cardinal News, SCOTUSblog, May 2026). * So Virginia runs November 2026 on its 2021 court-ordered maps, a 6-5 Democratic delegation, while Texas runs on a redraw worth roughly five Republican seats that the US Supreme Court left standing on April 27, 2026. Two states, two attempts to redraw mid-decade, opposite outcomes — one blocked over the calendar on which a legislature scheduled a vote. That asymmetry, not the generic ballot, is where this election is actually being decided. * **What is on the ballot.** Tuesday, November 3, 2026. Every one of the 435 seats in the House of Representatives, because the whole House is elected every two years. In the Senate, 33 seats on the regular six-year rotation plus two special elections, for 35 races. Also on the ballot in most places: governorships, state legislatures, attorneys general, ballot measures, school boards, judges. The federal races get the coverage; the state ones increasingly decide more about your daily life. * **The House arithmetic.** Republicans currently hold the chamber 218-214, with three vacancies. That is about as thin as a working majority gets. Democrats need a net gain of three districts to take control. Republicans can afford to lose two. * For context on how normal a flip would be: the president's party has lost House seats in the overwhelming majority of midterms since the Second World War. Midterms are structurally a referendum on the incumbent administration, and the incumbent administration usually loses ground. That is the base rate a Republican defence is fighting against, before any specific issue is factored in. * **The Senate arithmetic is different, and much harder for Democrats.** Republicans hold 53 seats. Democrats hold 45, and two independents caucus with them, giving them 47 votes for organisational purposes. To take the chamber, Democrats need a net gain of four. Republicans need only to hold their losses to two. * The exposure is more favourable to Democrats than it has been: of the 33 regularly scheduled races, Republicans are defending 20 seats and Democrats 13. But Senate seats are statewide, and statewide contests do not bend to national mood as easily as individual districts do. A party can win the national House vote comfortably and still fall short in the Senate, because the Senate is a map of states, not of people. * **Now the part most coverage buries: the maps.** * American congressional districts are normally redrawn once a decade, after the census. Redrawing them mid-decade, for purely political reasons, was until very recently something that essentially did not happen. Since the 1960s, exactly one state had done it voluntarily: Texas, in 2003. * This cycle broke that. As of March 2026, six states had enacted entirely new congressional maps for the November election — Texas, California, Missouri, North Carolina, Ohio and Utah. * Texas moved first, with a redraw estimated to be worth about five additional Republican-leaning seats. It was challenged, and on April 27, 2026 the Supreme Court overruled a lower court and left the map standing through at least 2030. * California responded in kind. In November 2025 its voters approved a Democratic counter-map, explicitly framed as a response to Texas, worth roughly five additional Democratic-leaning seats for this cycle. Missouri's legislature passed a new map in a September 2025 special session, signed by Governor Mike Kehoe, adding about one Republican district to its eight-seat delegation. Ohio's redistricting commission approved a map on October 31, 2025 that converts a 10-5 Republican advantage into 12 Republican-leaning districts and 3 Democratic-leaning ones. * Add it up and something uncomfortable emerges. A large share of the 2026 House outcome was decided in state capitols and courtrooms in 2025 and early 2026, before a single voter did anything. The maps are the campaign. * **The ruling that made the map wars easier.** On April 29, 2026, the Supreme Court decided Louisiana v. Callais by 6-3, in an opinion by Justice Samuel Alito. The decision substantially raises the bar for bringing a claim under Section 2 of the Voting Rights Act — the provision that bars voting rules and district lines that discriminate on the basis of race — including by demanding proof of intentional discrimination. Voting-rights organisations including the Campaign Legal Center, the NAACP Legal Defense Fund and the Center for American Progress describe the result as leaving Section 2 close to unusable in practice. * Whatever you think of the reasoning, the practical consequence is not in dispute: the main federal legal check on racially discriminatory map-drawing got much weaker, in the middle of the most aggressive map-drawing season in sixty years. * **How to read the polling without being fooled.** The generic congressional ballot — "would you vote for the Democrat or the Republican in your district?" — is the single most useful number in midterm coverage, and the single most abused. * As of early August 2026, Democrats lead it. By how much depends on who you ask: Morning Consult has 46-42, a YouGov/Economist poll fielded August 1-4, 2026 has 43.7-38.4, Decision Desk HQ's average runs around 46-40, and US Polling Data's average has 48.1-41.1. Data for Progress had it at D+8 on May 21, 2026. * That is a spread from about D+4 to about D+8 for the same election in the same fortnight. The gap comes from different pollsters, different likely-voter screens and different averaging methods — not from voters changing their minds four times a week. If a headline gives you one number with no source and no range, it is telling you less than it appears to. * And the generic ballot measures votes, not seats. Because of how districts are drawn — including all of the above — a party generally needs to win the national House vote by a clear margin, not a fraction of a point, to win a majority of the seats. A tied popular vote is not a tied chamber. * **What we are watching next.** Primary calendars through the summer and autumn; any further mid-decade map changes or court rulings on the ones already enacted; candidate filing and retirements, which are one of the most reliable early signals a party expects a bad year; special elections, which often swing harder than national polling and get read as omens; and the generic ballot averages, which matter far more as a trend line than as any single week's number.