While it’s not ironclad, there’s an adage in American politics reliable enough to qualify as conventional wisdom: The president’s party loses seats in the midterms.
For Republicans looking toward November, the history is not comforting. The president’s party has lost House seats in 20 of the 22 midterms since 1938. The exceptions, Bill Clinton in 1998 and George W. Bush in 2002, came under circumstances unusual enough to prove the rule, or at least give it legitimacy.
So history offers Republicans a warning.
What it does not offer, however, is a particularly specific forecast.
The number that matters (or does it?)
Since 1982, presidents facing the first midterm after taking office have seen their parties lose an average of roughly 28 House seats.
That sounds ominous until one looks underneath the average.
Barack Obama’s Democrats lost 63 seats in 2010. Clinton lost 54 in 1994. Donald Trump’s Republicans lost 41 in 2018.
But Joe Biden’s Democrats lost only nine in 2022. George H.W. Bush’s Republicans lost eight in 1990. George W. Bush actually gained eight in 2002.
In other words, the historical rule predicts direction far better than magnitude.
Of course, House Republicans entered this cycle with only 220 House seats. They do not need anything resembling 1994, 2010, or 2018 to lose the majority. A relatively modest midterm correction could do it.
That makes the historical average both alarming and a bit beside the point.
Which Trump midterm is this?
There is another challenge in comparing the 2026 cycle with history: how to categorize this Trump administration.
This is the first midterm of a newly elected Republican administration. It is also the second midterm of Trump’s presidency. His two terms just happen to be separated by four years and another president.
Those categories historically produce different results.
In first-term presidential midterms since 1982, the party in the White House has lost about 28 House seats on average. In the four conventional second-term midterms during that period, the average loss was closer to 11.
Trump fits both descriptions, but not neatly.
There’s one true precedent for a president serving nonconsecutive terms: Grover Cleveland. It is not especially reassuring for the GOP. In 1894, during Cleveland’s second, nonconsecutive term, Cleveland’s Democrats suffered the largest House wipeout in American history, dropping from 218 seats to 93.
But given that the election was held amid the Panic of 1893, the Pullman Strike, and an entirely different party system, Cleveland’s example is more of a historical curiosity than a usable model.
The House has changed, too
The larger reason to be cautious with historical comparisons is that today’s House is simply less elastic, with fewer districts that routinely swing between the parties.
Partisan sorting has packed Democratic and Republican voters into increasingly predictable seats. And this year’s mid-decade redistricting fight has further altered the battlefield.
The Cook Political Report’s latest analysis estimates that Republicans could net around four seats from the mid-decade map changes alone. Cook currently rates 21 House races as Toss Ups, with 16 of them held by Republicans.
The playing field is also remarkably compact. Cook notes that 86 percent of Republican-held seats are in districts Trump carried by more than 10 points in 2024. Its September battleground poll found Democrats leading 49 percent to 47 percent across the 37 districts it considers competitive, even though Trump won those districts by an average of five points two years ago.
In other words, the GOP may be better insulated from a 40-seat wipeout than governing parties were in previous eras.
Unfortunately for House Republicans, they also have much less room for error.
The signal to watch
None of this means history should be ignored.
The underlying conditions still look considerably midterm-ish. A Reuters/Ipsos poll released this week put Trump’s job approval at 32 percent, while just 17 percent approved of his handling of the cost of living. Among registered voters, Democrats led Republicans 43 percent to 35 percent when asked how they planned to vote in the midterms, although national congressional polling cannot capture the district-by-district contests that determine control of the House.
That looks quite different from the circumstances surrounding the two modern exceptions to the midterm rule. Clinton’s approval stood around 65 percent shortly before the 1998 election, while George W. Bush was at 67 percent in late October 2002.
So by the traditional measures, 2026 looks much more like a conventional midterm environment than one of the rare exceptions.
The more interesting question is whether Republicans will suffer the kind of losses that have traditionally accompanied an unpopular president’s first midterm. A more polarized electorate, fewer competitive districts, and favorable redistricting could spare Republicans anything resembling the 63-seat Democratic rout of 2010 or the 41-seat Republican loss of 2018.
That may be a small comfort.
With such a narrow majority, Republicans could outperform the historical averages considerably and still lose control of the House. And they may yet claim a victory of sorts over the midterm curse.
It would be a decidedly Pyrrhic one.