Why Does the President’s Party Almost Always Lose Seats in the Midterms?

Midterm season just kicked off, and you are about to hear nonstop that history favors whichever party does not hold the White House. Here is what that pattern actually looks like in the numbers, and why it keeps happening.

Now that Labor Day has passed, the 2026 midterm campaign is officially in full swing, and you are going to hear one statistic repeated constantly between now and November: history is not kind to whichever party holds the White House. It gets said so often it starts to sound like a slogan rather than a fact. It is worth actually looking at the numbers behind it, because the pattern is real, remarkably consistent, and only broken a handful of times in nearly a century.

The short answer: since 1934, the president’s party has lost House seats in almost every single midterm election, averaging a loss of about 28 seats, with only three exceptions in the modern era.

The numbers behind the pattern

Going back to 1930, the president’s party has lost an average of roughly 28 House seats and about 4 Senate seats in each midterm election. The losses are not always modest either. In 2010, during the Tea Party wave against President Obama, Democrats lost 63 House seats in a single night, one of the largest midterm swings on record. On the other end, the best a president’s party has typically managed in a midterm is a small gain, and even that has happened only rarely.

The three times the pattern broke

Three midterms in the modern era stand out precisely because the president’s party gained House seats instead of losing them, and each has its own story. In 1934, Franklin Roosevelt’s Democrats gained 9 House seats in the middle of the Great Depression, riding strong public approval for the New Deal at a moment when Roosevelt’s popularity was unusually high. In 1998, Bill Clinton’s Democrats gained 5 House seats even as Republicans pursued his impeachment, a result widely read at the time as public backlash against what many voters saw as Republican overreach, though some analysts have since argued the race was shaped more by an unusually strong economy than by impeachment politics specifically. In 2002, George W. Bush’s Republicans gained 8 House seats, propelled by a rare surge of national unity and high presidential approval in the year following the September 11 attacks. Each exception lines up with a president’s approval rating pushing unusually high, above 60%, at the moment voters went to the polls.

Why this pattern exists in the first place

Researchers point to a few overlapping explanations rather than one single cause. Turnout composition is a big one: presidential elections draw a larger, more casual electorate, with turnout often around 60%, while midterm turnout drops closer to 40% and skews toward more consistently partisan voters. That shift removes a chunk of the casual support that helped elect the president in the first place. Presidential approval also tends to decline over time in office, simply because governing forces choices that alienate some part of the coalition that elected a president, and that decline drags down congressional candidates who share the president’s party label. A Duke University study of the 2010 midterms even found a kind of blame asymmetry at work: a large share of voters who said both parties were equally responsible for the country’s problems still ended up voting against the president’s party by a wide margin, a dynamic researchers estimated could swing outcomes by several percentage points on its own.

The pattern is not destiny, though

The 2022 midterms are a useful reminder that this historical pattern describes a tendency, not a guarantee. Joe Biden’s approval rating sat around 42% heading into that election, a level that historical models would have predicted a substantial House loss for Democrats, in line with 2010 or worse. Instead, Democrats lost only 9 seats, a much smaller loss than the pattern alone would suggest. Analysts pointed to a mix of factors working against the usual trend that year: new congressional maps drawn after the 2020 census, weaker than usual Republican candidates in several key races, and a surge in turnout tied to abortion policy following the Supreme Court’s Dobbs decision earlier that year.

Why this matters heading into 2026

None of this is a prediction about what will happen this November, and the 2022 cycle is a real warning against treating the historical average as destiny. What the pattern does tell you is the baseline any given midterm has to beat or fall short of. Republicans currently hold narrow majorities in both chambers heading into 2026, which means, historically speaking, they are the party facing the version of this pattern that has held in the vast majority of midterms since 1934. Whether 2026 ends up looking more like 2010, more like 2022, or more like one of the three genuine exceptions depends on exactly the kind of approval numbers, candidate quality, and turnout dynamics researchers point to as the real drivers, not on the calendar alone.