Because my previous column caused a bit of a stir, it seemed worth fleshing it out a bit more. The thesis was that things had changed rather dramatically in the midterm outlook from what might have been expected in early 2025, to the point where the three most likely scenarios for Republicans were bad, really bad, or truly horrible.
It would be an overstatement to call it a dilemma, but two contradictory things seem to be at work. There are unmistakable reasons to believe that Republicans would suffer a net loss of House and Senate seats. President Trump’s awful job-approval ratings heading into a midterm election make that virtually inevitable, but there are also persuasive reasons to think that the losses would be no worse than average, maybe even a bit less than that. Republicans, at this moment, have 220 House seats (including one vacancy plus the seat held by Rep. Kevin Kiley, technically an independent who caucuses with Republicans) to Democrats’ 215; this is about as lean a majority as you would ever find, far narrower than the 295 seats that Democrats had going into the 1966 midterm under President Lyndon Johnson, or even the 225 seats that Democrats had under Bill Clinton in 1994 before that debacle or the 224 they held going into Barack Obama’s disastrous 2010 midterm. Republicans had 241 seats going into Trump’s 2018 midterm and 232 before George W. Bush’s 2006 midterm loss. As this column has noted before, you can’t get blood out of a turnip: A party cannot lose a seat it doesn’t have.
The best predictor of the outcome of any senatorial, congressional, or gubernatorial general election is the outcome in the previous presidential election. Republicans entered this Senate election with two seats that Trump won in 2024 by at least 40 points, six that he won by 30 to 39 points, five that he won by 20 to 29, and seven that he won by 10 to 19, meaning that Trump carried 20 out of 22 GOP seats by double-digit margins. The only exceptions are Maine, where Trump lost by 7 points, and North Carolina, where he won by 3. That would seem to severely limit Democratic opportunities to capture GOP-held seats.
But that’s the theory; let’s now talk about the reality of Sept. 17, 2026. Presidents with an approval rating as low as Trump’s are going to have midterm-election problems, and for several months now, all of the lights have been flashing “danger, danger, danger” for the GOP.
Sifting through all of the available public polls—we are talking hundreds here since Memorial Day—21 very high-quality polls in key Senate and/or governor battleground states have been released: seven by the New York Times/Siena, eight for Fox News, four commissioned by AARP, and last week, CNN released its first of hopefully a large number of statewide polls, in Maine and Michigan, both with key Senate and governor races. Both the Fox and AARP polls (of all likely voters, not just seniors) were conducted by bipartisan polling teams, while the NYT/Siena and CNN polls were done by nonpartisan pollsters. Trump’s approval ratings in these surveys have ranged from as low as 31 percent (CNN in Maine) to as high as 48 percent in the New York Times poll back in June. His approval rating has ranged from as low as 7 points below his 2024 vote share to as many as 17 points below in the June New York Times/Siena poll.
Suffice it to say that the president is a major liability for GOP Senate candidates. By my count, nine Senate races are now genuinely competitive: six in GOP hands (Alaska, Iowa, Kansas, Maine, Ohio, and Texas) and three held by Democrats (Michigan, Minnesota, and New Hampshire). Only four of the nine have incumbents still in the hunt, all Republicans: Sens. Dan Sullivan of Alaska, Roger Marshall of Kansas, Susan Collins of Maine, and Jon Husted of Ohio. All three Democratic seats on this very competitive list are open. Out of all nine, the one real surprise to me is Kansas; I was dismissive of such talk until the last few weeks.
Arithmetically speaking—and I emphasize this is purely theoretical—the range of outcomes could be anywhere from a net loss by Democrats of three seats to a net loss for Republicans of six. The two extreme outcomes are almost impossible, but it does make the point that there are nine very tight races and the range of outcomes is mind-boggling. It should be remembered that in the end, the tightest races rarely split down the middle—half going to one party, the other in the opposite direction. The breaks tend to go overwhelmingly in one direction.
Twenty of those blue-chip polls asked about Senate races. The Democratic candidate led in 15 surveys, the Republican in four, with one poll showing a tie. In 19 governor-race polls, Republicans led in two: an Ohio Fox poll from June and another Fox poll in Texas from July.
Almost all of these leads were tiny, generally within the margin of error. The point of this is not to say that one side is doing really well in the polling, but the polling is showing some extremely close races, ones that could break in either direction.
Generally speaking, with the exception of Texas and maybe Kansas, GOP problems are almost all Trump-related. In those vulnerable GOP-held seats, a substantial share of the electorate live in rural areas and small towns. They’re among the most negatively impacted by tariffs and the various trade wars the administration is engaged in, to say nothing of high gasoline, diesel, and fertilizer costs. In Alaska, Iowa, Ohio, and Texas, Republicans are struggling in both Senate and governor’s races, and all four states have long been dominated by the GOP. Ken Paxton of Texas and Sen. Marshall of Kansas have their own problems that exacerbate the political atmosphere.
For Democrats, in two of the three states that they should be most worried about, Maine and Michigan, they have suffered self-inflicted injuries. The other, the open seat in New Hampshire, should not be that much of a surprise. The Granite State has elected Republican governors in five consecutive elections, along with pretty consistent GOP majorities in both state legislative chambers. The GOP also has a solid candidate in former Sen. John E. Sununu. Granite State voters know the difference between a Sununu and a Trump, diminishing the presidential undertow in that state a bit.