The United States presidential election seems to have settled into a sort of trench warfare, with former President Donald Trump and Vice President Kamala Harris separated by a few points in national polls, and by a hair’s width in key swing states. Things could break out one way or the other in the next few weeks, but for now, a photo finish (absent some sort of system-wide polling error) seems likely.

Comparatively little attention has been paid to the battle for the Senate. This isn’t surprising: Presidential elections always receive top billing, and Senate races tend to coalesce fairly late in the game. What conventional wisdom there is seems to have congealed at Republicans winning 51 to 52 seats in the Upper Chamber.

This is probably right, as a best guess. Republicans hold 49 seats, and everyone seems to agree that West Virginia is going to go to a Republican, giving Republicans both West Virginia Senate seats for the first time since the 1950s and full control of the delegation for the first time since the 1920s. Sen. Jon Tester of Montana is behind in polls, which is a bad position for any incumbent to find themselves in.

Sen. Sherrod Brown of Ohio is in better shape against Bernie Moreno, but is only up about 4 points in the polls in a state where he has to convince one in 10 Trump voters to also vote Democrat for Senate. There are also two Republican senators – Ted Cruz of Texas and Rick Scott of Florida – who aren’t polling particularly well, but most view their seats as relatively safe absent a massive poll error.

So 51-52 is a fairly safe bet. What’s gotten less attention, though, is that if we look at the range of realistic possibilities, the Democratic tail is fairly short. It’s possible for them to pick up a seat or break even by defeating Cruz and Scott and holding on to all their seats save West Virginia and perhaps Montana.

What’s gotten even less attention is that the tail for Republicans is quite large. Let’s start with a simple fact: In 2016 and 2020, there was exactly one Senate race where a candidate from a party that didn’t win the state’s electors emerged victorious: Susan Collins. To be clear: Susan Collins is a political freak, and it isn’t entirely clear that she’ll be able to win reelection in 2026 (if she wants to). There aren’t many Susan Collinses in the world, and the two closest analogues are either retiring (Manchin) or likely to lose (Tester).

There are five senators, all of whom are Democrats, from states genuinely considered “in play” for the presidential election. If they all lose – which I don’t expect – Republicans could win 57 seats. The good news for Democrats is that they’re leading in every state. The bad news is that there are cracks in the walls for all of these candidates.

Let’s start with Pennsylvania, which is generally viewed as the most likely state to give the presidential winner the 270th electoral vote. Sen. Bob Casey currently leads Dave McCormick by 4 percentage points in the latest RealClearPolitics Average, and the race may be tightening as the two most recent polls have it closer. But let’s compare the presidential and Senate numbers from several recent polls. Susquehanna has the presidential race tied 46-46. It has Casey up 48-40. Muhlenberg shows a 48-48 presidential tie, with Casey up 48-43. Emerson has Trump up 48-47, and shows Casey leading 47-42.

Overall, the RCP Average has the presidential race in Pennsylvania as a statistical tie (Trump actually leads 48.1% to 47.9% for Harris with Casey at 48.4% and McCormick at 44.4%). Notice anything? There are probably a great many things to notice, but what jumps out at me is this: In all of these polls, Casey is getting nearly the same vote share as Kamala Harris. Casey may well be up 4 points right now. But the undecided voters are disproportionately likely to be Trump voters.

If Trump wins the state, that’s a problem for Casey. It isn’t that there won’t be any ticket splitting – of course there will be, and Casey may be the beneficiary. It’s more that Casey’s lead is probably an illusion. There’s a large chunk of the electorate that is committed to voting for Donald Trump, yet professes to being unsure about whether they’re voting for McCormick. Maybe they won’t, but I have my doubts.

In Michigan, Kamala Harris leads Donald Trump 48.9-47.5, and Elissa Slotkin leads Mike Rogers 47.6-43.2. In Wisconsin, Harris leads 49.1-48.3, while Tammy Baldwin leads Eric Hovde 49.7-46.1. In Arizona, Trump leads 49.3 with 47.3 for Harris – as Ruben Gallego leads Kari Lake 49.3-43.5. In Nevada, Harris leads 48.6-47.3, while Jacky Rosen leads Sam Brown 49.3-40.7.

What does this mean? Democrats lead in all of these Senate races, sometimes by comfortable margins. But those margins are fragile. The undecided voters are disproportionately Trump voters, which suggests these races are going to close in the next few weeks.

Again, many of these Democrats will run ahead of Harris. If Harris wins, they’ll probably be fine. But I don’t think they’ll likely run more than a couple of points ahead of Harris. If Trump is winning the Electoral College, it isn’t fanciful for most of these seats to also flip, setting up a GOP Senate that looks very, very different than it did during Trump’s first term.