Politics

What the US Midterm Elections Decide and How Often the President’s Party Loses Seats

Midterm elections decide control of Congress. Here is what history shows, where the 2026 House and Senate tallies stand, and what forecasters say.

By The BB Report October 29, 2026

What the US Midterm Elections Decide and How Often the President’s Party Loses Seats — The US Capitol in Washington, where the House and Senate sit. Photo: Noclip / Wikimedia Commons (Public domain)

Key takeaways

Midterm elections decide control of Congress, and divided government usually limits a president’s ability to launch major new policy, the Presidency Project says.
Since 1946 the president’s party has lost 25 House seats on average, according to Gallup, with larger losses when approval is below 50%.
For 2026 the Silver Bulletin describes Democrats as strong favourites to retake the House, while the Presidency Project’s regression projects large Republican losses in both chambers, with large error bounds.

US midterm elections decide who controls Congress, and the Presidency Project says divided government usually limits a president’s ability to launch major new policy.

The pattern is familiar: the president’s party almost always loses House seats in midterm elections. The Presidency Project says how much it loses depends on approval ratings and on how many seats the party has to defend. This explainer sets out what the 2026 midterms decide, what history shows and how analysts read the current race, with the figures attributed to the people who published them.

What Is on the Ballot

Members of the House of Representatives serve two-year terms. The House’s official website says the number of voting representatives is fixed by law at no more than 435. It lists two vacancies at present, Cherfilus-McCormick’s seat in Florida’s 20th District and Gonzales’s seat in Texas’s 23rd District.

The starting tally differs a little by source. Brookings nonresident senior fellow David C. Barker wrote on 8 September that Democrats start five seats down, at 214 to 219 for Republicans, with two vacancies in play. "To regain the gavel, then, the Democrats need just three net wins." The Presidency Project counts Republicans at 218 seats, not 219, which it describes as precisely a majority, and says Kevin Kiley, an independent who caucuses with them, is not a reliable vote.

In the Senate, the Presidency Project counts Republicans at 53 seats, two of them vacant, and Democrats at 45, usually joined on important issues by independents Bernie Sanders and Angus King. Twenty seats are up on the regular calendar, plus the two vacancies, so Republicans defend 22. Democrats would need to gain four Senate seats for an effective working majority, the same analysis says.

Midterm Numbers to Know

The House has no more than 435 voting members, fixed by law, according to the House’s official website. Since 1946 the president’s party has lost 25 House seats on average, 37 when approval is below 50% and 14 when it is above, according to Gallup. In 2026 Democrats need a net three House seats, per Brookings, and four Senate seats for an effective working majority, per the Presidency Project.

What a Change of Control Does

The Presidency Project’s 2022 analysis said a president’s ability to launch major new initiatives "would be greatly weakened under divided government". Its 2026 analysis says divided government usually limits a president’s ability to launch major new policy.

If Republicans lose control of either house, the Administration can expect to be targeted with disruptive and aggressive congressional inquiries, the 2026 analysis says.

What History Shows

Gallup says that for all midterm elections since 1946, the president’s party has averaged a loss of 25 U.S. House seats. Gallup’s table shows that presidents with approval below 50% have seen their party lose 37 House seats on average, against an average loss of 14 seats when approval was above 50%. Gallup describes the link between approval and seat loss as strong, with a correlation of.66.

Only two presidents saw their party gain seats in midterms since 1946: Bill Clinton in 1998, up five seats at 66% approval, and George W. Bush in 2002, up six at 63%. Gallup’s table also shows the range: Harry Truman’s party lost 55 seats in 1946 at 33% approval, while Jimmy Carter’s lost 11 in 1978 at 49%.

The Presidency Project’s longer series, covering 1934 to 2022, puts the average loss at 27 House seats and three Senate seats. In that period the president’s party gained House seats only three times, and lost seats in both chambers about two-thirds of the time. Barker’s Brookings analysis finds that in only three midterms since the Civil War has the president’s party not lost House seats: 1934, 1998 and 2002. The two analyses ask different questions, so their counts are not directly comparable.

To regain the gavel, then, the Democrats need just three net wins.

How Forecasters Read 2026

On 29 October the Silver Bulletin’s FLIPR model described Democrats as strong favourites to retake the House, with their Senate odds holding steady. It listed Democratic polling leads in Senate battlegrounds of 2.5 points in Texas, 3.0 in Ohio, 3.6 in Alaska and 2.4 in Maine, and cautioned that these leads are "just a normal-sized polling error" away from flipping. The topline numbers are typically based on 40,000 simulations, and the model can estimate combined outcomes, such as Republicans holding the Senate but losing the House.

Barker wrote on 8 September that Democrats are positioned to gain around 25 to 30 House seats, more than the net three they need but short of a landslide. He cited generic-ballot polls showing Democrats ahead by about six points, and presidential approval in the mid-to-high 30s, which is among the lowest of any president at this point in his term since 1946. He wrote that the smart money puts the Democratic gain at around 17 seats, and said he would take the over.

The Presidency Project’s regression produces a larger figure, around 43 House seats and 12 Senate seats lost by Republicans, with large error bounds. It describes Republican losses of fewer than four Senate seats or 25 House seats as unexpectedly positive for the party, and cites Trump’s latest approval at about 37%. Its model has no measure of current national controversies, such as the war in Iran, tariffs and Trump-related court fights, or of candidate quality.

Reasons for Caution

Barker lists several counterweights. Republican committees and allied groups currently hold a half-billion-dollar cash edge, but in the second quarter of 2026 Democratic challengers outraised Republican incumbents in about two-thirds of the toss-up districts. Republican redistricting moved the median district about two points toward Republicans, and Barker says that damage is already in his count. He writes that only about three or four dozen districts are truly competitive.

Polling is another unknown. Barker notes that swing-state polls overestimated Democratic support in 2016, 2020 and 2024, while polls in 2018 and 2022 were relatively accurate, and in 2022 underestimated Democrats. Democrats still lost nine House seats in 2022 despite President Joe Biden’s unpopularity. Barker writes that a meaningful share of Biden’s disapprovers were frustrated Democrats who were never going to vote red, and that relatively few of Trump’s disapprovers are Republican voters.

Turnout is another variable. Barker writes that unpopular presidents tend to dampen enthusiasm among supporters and boost it among opponents, and the Presidency Project says turnout is likely to be larger than normal for a midterm election, given the intensity of current controversies.