The president’s party usually loses seats in midterms because the voters who are unhappy with the president are more motivated to turn out than the voters who are satisfied, and because some swing voters use the midterm to check the party in power. Since 1934, the president’s party has gained House seats in a midterm only three times: 1934, 1998 and 2002. Over the 20 midterms from 1946 through 2022, it lost an average of about 26 House seats and between three and four Senate seats.
How big the loss gets tracks the president’s approval rating closely, though not perfectly. Here’s the record, the main explanations, and what to watch before November 3, 2026.
What does the midterm record look like? #
Here are recent midterms, with the president’s late-October approval and the seat changes for his party, from the American Presidency Project.
| Year | President | Approval | House seats | Senate seats |
|---|---|---|---|---|
| 1994 | Bill Clinton (D) | 48% | −54 | −8 |
| 1998 | Bill Clinton (D) | 65% | +4 | 0 |
| 2002 | George W. Bush (R) | 67% | +8 | +2 |
| 2006 | George W. Bush (R) | 37% | −30 | −6 |
| 2010 | Barack Obama (D) | 45% | −63 | −6 |
| 2014 | Barack Obama (D) | 41% | −13 | −9 |
| 2018 | Donald Trump (R) | 44% | −41 | +2 |
| 2022 | Joe Biden (D) | 40% | −9 | +1 |
The two gains in that span came with approval in the mid-60s: 1998, amid a backlash to the Republican-led impeachment effort against Clinton, and 2002, a year after the September 11 attacks.
Why does the president’s party lose ground? #
Political scientists have offered several explanations. They aren’t mutually exclusive.
Surge and decline #
Angus Campbell described this in 1960. A presidential election draws a “surge” of occasional voters who are pulled in by the big race, and they tend to favor the winner, helping his party’s congressional candidates. Two years later there’s no presidential race, those voters stay home, and the party loses the seats they helped win. We cover the turnout side in why voter turnout is lower in midterms.
Negative voting #
Samuel Kernell argued that dissatisfaction motivates more than satisfaction. Once a president governs, every decision disappoints someone, and those disappointed voters are the most eager to vote in the midterm. Supporters who got what they wanted feel less urgency.
Balancing #
Some voters, especially independents, prefer divided government. Alberto Alesina and Howard Rosenthal developed the idea that moderates use the midterm to pull policy back toward the center by voting against the president’s party.
Exposure #
The more seats a party won in the previous election, the more marginal seats it has to defend. A big presidential win sweeps in members from districts that normally lean the other way, and those are the first to fall two years later.
How does approval affect midterm losses? #
Approval is the single best indicator of how bad a midterm will be. In the American Presidency Project’s table, 16 midterms since 1938 have a late-October approval figure. In the five where the president was at 50% or higher, his party lost an average of about 14 House seats. In the 11 where he was below 50%, it lost about 35.
The relationship isn’t mechanical. In 1938, Franklin Roosevelt was at 60% and Democrats still lost 72 House seats. In 2022, Joe Biden was at 40%, but Democrats lost only 9 House seats and gained a Senate seat, a result many analysts linked to the Supreme Court’s abortion ruling that summer and to weak candidates in some races. Approval sets the baseline; the specific year can move the result around it.
That makes approval worth watching closely through October. Election Tracker shows presidential approval as a 7-day average of every published poll, with approve, disapprove, the net number, and whether it’s rising, falling or stable, plus a 90-day chart and the individual polls beneath it. We explain how to read those numbers in how to track presidential approval ratings.
Why does the Senate behave differently? #
Only about a third of Senate seats are up in any year, and which states happen to be voting matters as much as the national mood. In 2018, Republicans lost 41 House seats but gained two Senate seats, because Democrats were defending many seats in states Trump had won. The House, with all 435 seats up, reflects the national environment more directly.
What about the 2026 midterms? #
In 2026 the president’s party is Republican, with Donald Trump in his second term. Several factors point in different directions, and none is a forecast:
- History says the president’s party usually loses House seats, with the size of the loss tied to approval.
- Redistricting changed district lines in a number of states in 2025 and 2026, so the relationship between votes and seats won’t match 2024. We cover that in what the generic ballot is and how to read it.
- The Senate map has Republicans defending 22 seats to Democrats’ 13, but many of those Republican seats are in states the party usually wins easily. See which Senate seats are up in 2026.
The best way to follow it is through the numbers that historically matter: presidential approval and the generic congressional ballot. Election Tracker’s House tab has the generic ballot’s 30-day average and trend line, and its Market Sentiment tab shows current Polymarket prices on House and Senate control, labeled as trader sentiment rather than forecasts.
Frequently asked questions #
Why does the president’s party lose more House seats than Senate seats? #
All 435 House seats are up every two years, so the whole chamber is exposed to the national mood. Only about a third of the Senate is up, and the particular states voting that year can favor either party regardless of the national climate.
When has the president’s party gained seats in a midterm? #
Since 1934, three times in the House: 1934 under Franklin Roosevelt, 1998 under Bill Clinton, and 2002 under George W. Bush. All three came in unusual circumstances: the New Deal’s early popularity, a backlash to impeachment, and the aftermath of September 11.
How many seats does the president’s party usually lose? #
From 1946 through 2022, the average was about 26 House seats and between three and four Senate seats. The range is wide, from gains to losses of more than 60 House seats in 2010.
Is the midterm penalty getting smaller? #
Not clearly. Losses were large in 2006, 2010 and 2018, and small in 2002 and 2022. With fewer swing districts, there are fewer seats that can change hands, which may cap the size of future waves, but a low-approval president can still lose dozens of seats.
How do prediction markets treat the midterm penalty? #
Traders factor in the historical pattern along with polls, approval and the specific map. Market prices are sentiment from people with money at stake, not forecasts, and they move with the same news as the polls. Our comparison of prediction markets and polls covers how they differ.