The approval tracker is a daily weighted average of published approval polls.

How it differs from the global standard

• Approval, not vote intention: the figures are approve / disapprove of the president, not party vote share.
• Field-cluster benchmark: the estimate is judged against the full field of major aggregators rather than any single one. The field average and median are the trust benchmark; no one aggregator is treated as ground truth.
• One source: polls come from the New York Times file of approval polls. Polls from sources retired earlier are kept for the dates before their retirement.

Method in use: us-approval-2026.1

Methodology updated 17 October 2026. We regularly refine the method. The method is now: the average combines published national polls of Donald Trump's performance as president, from the New York Times file of approval polls. It includes polls of US adults and, where eligible, of registered or likely voters. It should be read as a summary of recent polling, not an exact measurement of approval among all US adults. Each release counts once towards a pollster's record, and only releases fielded by the day being calculated count. Where a poll reports more than one population, one figure is taken by the same order for every pollster. Reuters/Ipsos is weighted as a standard poll. We do not currently publish an uncertainty interval, and the 7-day and 30-day changes are shown once 7 and 30 days of figures under this method exist. Previously: polls were added to the record only within 45 days of fieldwork. Each population figure counted separately towards a pollster's record. Where two population figures from one poll were both held, the larger sample was used. Reuters/Ipsos was weighted as a tracking poll. The average was shown with a 95% uncertainty interval. Figures for earlier dates stay as they were stored on 17 October 2026.

• Source: the New York Times file of approval polls. Polls from sources retired earlier (Wikipedia from 16 May 2026; RealClearPolitics, pollster websites and 538 from 17 August 2026) count only if they were fielded and added before that date. Rasmussen Reports polls fielded after 28 July 2026 are not counted.
• What is included: polls with a fieldwork midpoint in the past 45 days and at least 400 respondents. Each pollster contributes its most recent release, and each release one population: the one the pollster leads with, then registered voters, then likely voters, then adults.
• Which pollsters are counted: a pollster with a quality score of at least 0.40, at least five releases and 90 days between its first and latest release. Pollsters scoring 0.70 or more need only the quality score.
• Weights: recency (a 14-day half-life), sample size, population and a fixed quality score for each pollster. Tracking polls carry a reduced weight, and a pollster counted on its quality score alone carries half weight until it has five releases.
• Uncertainty: no uncertainty interval is published at present.
• Revisions: each day the past 28 days of figures under this method are recalculated as late polls arrive. Figures for dates before 17 October 2026 stay as they were stored on that date.
• Change figures: the 7-day change is shown once 7 days of figures under this method exist, and the 30-day change once 30 days exist.
• Net approval: the published approve figure minus the published disapprove figure.

How to read it

It reflects the polls and the chosen field of sources; it does not second-guess them. Current figures, and every poll with its weight, are on the approval tracker.