The national landscape
What each of the 50 states and DC did to check its vote count after the 5 November 2024 General Election. Nine states ran two separate audits, so the table has more rows than states. Select any row for that state’s full summary and sources.
At a glance
Number of audits of each kind. A state with two audits is counted twice.
Every state
Rules as in force for November 2024. “If a discrepancy is found” is what the rules require, not what happened; red text marks an audit that could not change the result.
| State | Kind of audit | Races checked | If a discrepancy is found | When it ran |
|---|
What stands out
- Most audits are small, fixed-rate hand counts. The typical audit hand counts a set share of precincts or machines, often 1 to 5 percent, and compares the totals to the machine count. It can find errors, but it is not sized to tell whether an error was large enough to change who won.
- Only risk-limiting audits are designed to confirm the winner, and they check very few races. Nine states ran one. Each checked one to three races, usually President or a single statewide office. Every other race on those ballots was outside the audit’s guarantee.
- A strong audit can still be unable to change anything. Ohio has one of the most carefully specified audits in the country, and Wisconsin hand counted 327,230 ballots, nearly 10 percent of all its ballots. Both run after the results are certified, so neither can change the result it checks. How good an audit is and whether it matters are separate questions.
- States announce cleaner results than their records show. Rhode Island reported “no discrepancies”; its own data export showed 55. Illinois’s state log reported none, while Chicago lost ballots in two precincts and nine other jurisdictions filed differences. In Wyoming’s Weston County, an audit certified “no issues” and was later found to have been falsified or never performed.
- Results are hard to find, or never published. Rhode Island published 16 months after the election; Connecticut about a year. Maryland’s reports were taken offline within days. New Jersey’s are scanned images. Several states published nothing, and records requests were needed to learn what happened.
- Rules that require a response to a discrepancy often go unused. Missouri requires an investigation when a count differs by more than 0.5 percent; none is on record. Hawaii “shall” expand its audit; 9 of its 56 reports differed, and the state has no record of any expansion.
- Independent audits are rare. Most audits are run by the same office that ran the election. Only New Mexico and New Jersey use an outside party, and in Idaho university accounting students do the counting.
- Some audit laws let a political party stop the audit. In Arizona, two of 15 counties ran no hand count in 2024 because party chairs did not appoint the counting boards.
Where common sources get it wrong
National compilations are a good guide to what the laws say, and a poor guide to what ran in November 2024. Several describe laws passed after the election as if they applied to it.
| State | Often described as | What applied in November 2024 |
|---|---|---|
| Alabama | Hand-count audit law | No audit law in force. The 2021 law covered 2022 only; the current law was passed about 13 months after the election. |
| North Dakota | Hand-count audit | A re-run of the pre-election test ballots, not a check of voted ballots. The real audit law took effect in 2025. |
| Michigan | Procedural review only | A statewide risk-limiting audit of the presidential race, in all 83 counties. |
| Illinois | Hand count | Ballots from 5% of precincts re-run through the same kind of scanner; no one reads the ballots. |
| New York | Risk-limiting audit | A flat 3% hand count of voting machines. The bills to make it an RLA never passed. |
| Texas | “Hand Count Election Audit” | The older “partial manual count” (1% of precincts or 3). The new design took effect in September 2025. |
| Kentucky | 3–5% ballot audit, or the 2025 regulation | A hand-to-eye recount of one machine per county, new for 2024. The older rule was never carried out. |
| Wyoming | Hand count | Software comparing ballot images to the vendor’s records. The hand-count law applies from 2026. |
| Vermont | Risk-limiting audit | Every ballot in 7 towns re-scanned on separate equipment; no risk limit. |
| New Hampshire | Procedural review | Every ballot from 8 machines re-scanned on a different vendor’s scanner, first done in 2024. |
| Mississippi | Post-election audit | A review of procedures and paperwork in 4 counties. No votes are counted. |
Sources
Each state’s summary links to its own law and results. National compilations consulted: