President Donald Trump’s job approval stood at 31% in a Reuters/Ipsos poll released October 19, the lowest reading in that polling series across his two presidential terms. The survey also put Democrats ahead of Republicans, 47% to 35%, among registered voters asked which party’s congressional candidate they would support, Reuters reported.

The findings describe a difficult national environment for Republicans before the November 3 midterms. They do not establish how many seats either party will win. The approval result covers all adults, while the congressional preference result covers registered voters—two different populations that should not be treated as interchangeable.

What the new numbers show

According to Ipsos’s tables, 66% of adults disapproved of Trump’s performance. Among registered voters, his approval was 34%, not 31%. Those distinctions matter: an assessment of the president’s work is not the same question as choosing a congressional candidate, and neither response constitutes a recorded vote. Separately, 11% of registered voters said they did not know which party’s congressional candidate they would choose. That unresolved preference is another reason not to translate the party totals into a final election tally.

The same tables show that 47% of registered voters selected a candidate’s position on the cost of living as their most important consideration from the options offered. Democratic values and norms followed at 26%, with immigration at 12%. This was a question with specified response choices, not an unrestricted inventory of everything voters consider important.

The figures come from one Reuters/Ipsos study. Its published questionnaire and news coverage offer ways to check the findings, but they are not separate surveys independently reproducing the result. The poll measures stated attitudes; it cannot, by itself, establish why an individual changed an opinion or whether that person will vote.

How the survey was conducted

Ipsos interviewed 1,679 adults from October 16 through October 19 using its probability-based KnowledgePanel, according to its methodology statement. Interviews were conducted in English. The sample included 1,279 registered voters, and the published tables report a sampling margin of error of plus or minus 2.8 percentage points for that group.

For the full adult sample, Ipsos reports a margin of plus or minus 2.4 points at the 95% confidence level. The company weighted responses for demographic characteristics, geography, income, 2024 presidential vote and party identification. It identifies the benchmarks used and notes that rounding can make displayed percentages total slightly more or less than 100%.

These disclosures help readers assess what was measured and how. As the American Association for Public Opinion Research explains in its journalist guide, question wording, interview timing and the population being surveyed all affect interpretation. Weighting seeks to correct differences between respondents and the population; it does not turn a survey into a census of voters.

A record reading is not proof of a sudden shift

Reuters reported that the 31% approval estimate was one percentage point below its reading two weeks earlier. That is a new low within the series, but the numerical change alone does not demonstrate a statistically meaningful deterioration over those two weeks. Describing the record point estimate and establishing a reliable trend are different tasks.

Pew Research cautions that uncertainty around changes between polls can be larger than readers expect. The margin for an individual percentage is not automatically the margin for the difference between two findings. Looking for repeated movement across surveys provides a firmer basis for identifying a trend than emphasizing a small change between two measurements.

The 12-point congressional preference gap is more substantial than the one-point approval movement. Even so, a national party-preference question cannot be read as a projection of a particular district or state. AAPOR’s guidance is to treat election polling as a measurement of opinion at the time of interviewing, not as an election result.

The campaign continues in individual states

On the evening the poll was released, Trump appeared in Park City, Kansas, to campaign for Republican Sen. Roger Marshall, who faces Democratic pastor Adam Hamilton. KCUR reported that Trump urged supporters to treat the election as if he were on the ballot. Marshall emphasized priorities including border security, military strength and lower taxes.

Hamilton offered a different appeal, saying he would work with Trump when appropriate and oppose policies that harmed Kansas farmers and rural hospitals, according to Reuters’ coverage. The competing messages illustrate how candidates are presenting their cases locally. The national poll does not measure which argument will prevail in that Senate contest.

Turnout remains outside the headline

Registered voters are not identical to the people who ultimately cast ballots. Pew’s methodological guidance notes that even surveys restricted to likely voters depend on turnout models that introduce uncertainty. Sampling margins also do not capture every possible error, including differences in who responds or how a question is understood.

The immediate conclusion is narrower than a prediction of control of Congress: this survey records low presidential approval and a Democratic advantage in stated congressional preference. Whether those attitudes persist, translate into participation and produce victories in individual races remains unresolved. Subsequent polling and actual election returns will answer different questions from the one this study can answer today.