Election fraud statistics at a glance
Election fraud statistics are easy to overstate and hard to summarize well. The strongest pattern in the supplied data is not a single dramatic number, but a repeated mismatch between perception and volume: the federal record shows election-fraud matters as a tiny share of overall criminal caseloads, while state audit data and ballot review data show how election administration is usually checked in practice.
Table of contents
- Fast facts
- Federal prosecution data
- Where election-fraud cases were filed
- Statutes used in election-fraud charges
- Michigan audit and ballot review statistics
- Ballot rejection and drop box statistics
- What these statistics show
Fast facts
At a glance: the federal data in the dataset points to a narrow case category, not a large share of overall prosecutions.
- From FY2001-FY2017, DOJ Public Integrity Section filed 695 cases, and 19 were election-fraud related, about 3% of that caseload (GAO-19-485).
- In the same period, U.S. Attorneys’ Offices filed just over 1,000,000 criminal cases, and 185 were election-fraud related, or 0.02% of the total (GAO-19-485).
- Those 185 cases involved 309 unique defendants (GAO-19-485).
- U.S. Attorneys’ Offices initiated 525 election-fraud related matters from FY2001-FY2017 (GAO-19-485).
- At least one election-fraud related matter was initiated in 85 of 94 federal judicial districts (GAO-19-485).
- Michigan’s statewide audit review found no evidence of intentional misconduct or fraud by election officials (Michigan Post-Election Audit Report).
Federal prosecution data
Why the federal caseload matters
The cleanest way to read the federal data is as a denominator story. Election-fraud matters exist, but they sit inside much larger prosecutorial pipelines.
That matters because raw case counts can sound large in isolation. Once the totals are placed next to the full caseload, the scale changes:
- 19 election-fraud related cases out of 695 Public Integrity Section cases from FY2001-FY2017, about 3% (GAO-19-485).
- 185 election-fraud related cases out of just over 1,000,000 criminal cases filed by U.S. Attorneys’ Offices, or 0.02% (GAO-19-485).
Those figures do not erase the seriousness of any individual case. They do show that election-fraud cases are a very small slice of the federal criminal docket in the dataset.
Matter counts and defendants
The case and defendant counts add another layer of scale:
- The 19 Public Integrity Section election-fraud cases involved 37 individual defendants (GAO-19-485).
- The 185 U.S. Attorneys’ Office election-fraud cases involved 309 unique defendants (GAO-19-485).
- U.S. Attorneys’ Offices initiated 525 election-fraud related matters from FY2001-FY2017 (GAO-19-485).
A useful way to read those numbers is that matters, cases, and defendants are not the same unit. A small number of matters can involve more than one defendant, and a case can move through multiple charging decisions or legal theories.
Year-to-year concentration in the Public Integrity Section
The Public Integrity Section data shows that election-fraud cases were not evenly distributed across the years covered.
- 7 election-fraud cases were filed in FY2003 (GAO-19-485).
- 5 election-fraud cases were filed in FY2014 (GAO-19-485).
- The 19 election-fraud cases were part of a broader set of 695 Public Integrity Section cases from FY2001-FY2017 (GAO-19-485).
That pattern suggests periodic spikes rather than a constant stream. The supplied data does not explain the causes of those spikes, so the strongest defensible reading is simply that election-fraud filings clustered in particular years.
A compact comparison of the main federal totals
| Measure | Statistic | Source |
|---|---|---|
| Public Integrity Section cases, FY2001-FY2017 | 695 | GAO-19-485 |
| Public Integrity Section election-fraud cases | 19 | GAO-19-485 |
| Share of Public Integrity Section caseload | About 3% | GAO-19-485 |
| U.S. Attorneys’ Office criminal cases, FY2001-FY2017 | Just over 1,000,000 | GAO-19-485 |
| U.S. Attorneys’ Office election-fraud cases | 185 | GAO-19-485 |
| Share of criminal caseload | 0.02% | GAO-19-485 |
| U.S. Attorneys’ Office election-fraud matters | 525 | GAO-19-485 |
| Unique defendants in those cases | 309 | GAO-19-485 |
Where election-fraud cases were filed
District coverage was broad, but not uniform
The district-level numbers show wide geographic reach with strong concentration in a few places.
- At least one election-fraud related matter was initiated in 85 of 94 federal judicial districts (GAO-19-485).
- 49 of the 94 districts did not prosecute any election-fraud related cases from FY2001-FY2017 (GAO-19-485).
- Three districts accounted for 145 of 525 election-fraud matters, about 28% of the total (GAO-19-485).
That combination is important. Many districts saw at least one matter, but the workload was concentrated in a smaller set of districts.
The most active districts in the dataset
| District | Election-fraud cases | Share of all such cases | Source |
|---|---|---|---|
| Southern District of Florida | 42 | 23% | GAO-19-485 |
| Eastern District of Kentucky | 36 | 19% | GAO-19-485 |
| Eastern District of Wisconsin | 15 | 8% | GAO-19-485 |
Those three districts alone account for a meaningful share of the federal election-fraud case total in the dataset. The Southern District of Florida leads the list with 42 cases (GAO-19-485).
The long tail matters too
The concentration at the top does not mean the rest of the country was inactive.
- 50 of the 185 U.S. Attorneys’ Office election-fraud cases were filed in 42 other districts (GAO-19-485).
- 20 districts had only one election-fraud case during the period (GAO-19-485).
- 49 districts had no election-fraud related cases from FY2001-FY2017 (GAO-19-485).
That split creates a classic long-tail pattern: a few districts account for a disproportionate share of cases, while many districts have little or no activity.
Statutes used in election-fraud charges
About 100 different statutes were used
The charging data suggests variety rather than a single standard statute.
- U.S. Attorneys’ Offices used approximately 100 different statutes in election-fraud charges (GAO-19-485).
- 52 U.S.C. 10307 was charged in 52 election-fraud cases, making it the most frequently charged statute (GAO-19-485).
- 18 U.S.C. 371 was charged in 49 cases (GAO-19-485).
- 18 U.S.C. 1001 was charged in 39 cases (GAO-19-485).
- 52 U.S.C. 10307(c) was charged in 38 cases (GAO-19-485).
- 18 U.S.C. 611 was charged in 38 cases (GAO-19-485).
A compact view of the most frequently cited statutes helps show how spread out the charges were.
| Statute | Cases charged | Source |
|---|---|---|
| 52 U.S.C. 10307 | 52 | GAO-19-485 |
| 18 U.S.C. 371 | 49 | GAO-19-485 |
| 18 U.S.C. 1001 | 39 | GAO-19-485 |
| 52 U.S.C. 10307(c) | 38 | GAO-19-485 |
| 18 U.S.C. 611 | 38 | GAO-19-485 |
| 18 U.S.C. 911 | 21 | GAO-19-485 |
| 18 U.S.C. 1015(f) | 15 | GAO-19-485 |
| 52 U.S.C. 30122 | 15 | GAO-19-485 |
What the statute spread suggests
The statute mix is another signal that election-fraud cases are not one uniform category. The dataset shows multiple legal theories appearing across the charging record, with no single statute dominating every case.
The most defensible takeaway is simple: the legal framing of election-fraud allegations varies, and the federal record in the dataset reflects that variability (GAO-19-485).
Michigan audit and ballot review statistics
A statewide audit with a large sample size
The Michigan audit data gives a different kind of election-fraud statistic: not prosecution counts, but audit and review outcomes.
- More than 18,000 ballots were randomly selected from more than 1,300 local jurisdictions (Michigan Post-Election Audit Report).
- The audit selected 18,162 ballots for review (Michigan Post-Election Audit Report).
- The audit retrieved or randomly selected multiple times 18,804 ballots in total (Michigan Post-Election Audit Report).
- The audit produced 18,051 distinct ballots actually reviewed (Michigan Post-Election Audit Report).
- 21 Michigan local jurisdictions failed to retrieve ballots after multiple reminders (Michigan Post-Election Audit Report).
Those numbers show a broad audit footprint. The sample was large enough to examine statewide vote totals and local retrieval patterns without reducing the exercise to a small pilot.
Sampled ballots versus tabulated totals
The audit review gives two useful percentage comparisons for the presidential race.
- Biden had 49.7% of sampled ballots versus 50.3% in the tabulated total (Michigan Post-Election Audit Report).
- Trump had 48.0% of sampled ballots versus 47.5% in the tabulated total (Michigan Post-Election Audit Report).
Those differences are small. The sample tracked the tabulated totals closely enough that the gap is best understood as a narrow deviation inside a large statewide review, not a dramatic mismatch.
Michigan statewide presidential totals
| Measure | Votes | Share | Source |
|---|---|---|---|
| Biden statewide total | 2,804,040 | 50.3% of 5,579,317 ballots cast | Michigan Post-Election Audit Report |
| Trump statewide total | 2,649,852 | 47.5% of 5,579,317 ballots cast | Michigan Post-Election Audit Report |
| Biden among valid presidential ballots | - | 50.6% of 5,539,302 valid presidential ballots | Michigan Post-Election Audit Report |
| Trump among valid presidential ballots | - | 47.8% of 5,539,302 valid presidential ballots | Michigan Post-Election Audit Report |
County sample details
The county sample data shows where ballots were concentrated in the audit.
- Wayne County had 2,789 sampled ballots (Michigan Post-Election Audit Report).
- Oakland County had 2,484 sampled ballots (Michigan Post-Election Audit Report).
- Macomb County had 1,601 sampled ballots (Michigan Post-Election Audit Report).
- Kent County had 1,230 sampled ballots (Michigan Post-Election Audit Report).
The audit also notes that in the 20 Michigan counties with the highest number of 2020 voters, ballot retrieval ranged from 212 to 721 ballots per county (Michigan Post-Election Audit Report).
The strongest Michigan finding
The most important audit result in the supplied data is explicit:
- Michigan’s statewide audit found no evidence of intentional misconduct or fraud by election officials (Michigan Post-Election Audit Report).
That statement is narrower than a general claim about every possible kind of voting problem. It is also the clearest audit-level finding in the dataset, and it should sit at the center of any fair reading of the Michigan numbers.
Ballot rejection and drop box statistics
Rejected mail ballots in the 2022 EAVS
The EAVS figures help show why election administration statistics are often mixed: some categories are well defined, while others remain vague.
- 32.7% of rejected mail ballots were rejected for reasons not listed in the survey (EAVS 2022 Comprehensive Report).
- 26.9% were rejected for a non-matching or incomplete signature (EAVS 2022 Comprehensive Report).
- 23.1% were rejected because they were not received on time (EAVS 2022 Comprehensive Report).
These three figures already cover the bulk of the rejection pattern named in the dataset, and they show that a large share of rejected ballots may not be easy to categorize with precision.
Drop box reporting in 2022
The same EAVS data also includes drop box usage across states.
- 39 states reported using drop boxes (EAVS 2022 Comprehensive Report).
- The remaining 32 states reported 12,917 drop boxes total (EAVS 2022 Comprehensive Report).
- Of those, 11,858 were available on Election Day and 8,386 were available during early voting (EAVS 2022 Comprehensive Report).
This is a useful reminder that election administration statistics often mix policy, logistics, and reporting structure. Numbers about drop boxes are not election-fraud allegations by themselves, but they do matter because they describe how ballots are handled and how states report that infrastructure.
Quick comparison of the administrative statistics
| Measure | Statistic | Source |
|---|---|---|
| Rejected mail ballots for reasons not listed | 32.7% | EAVS 2022 Comprehensive Report |
| Rejected mail ballots for non-matching/incomplete signature | 26.9% | EAVS 2022 Comprehensive Report |
| Rejected mail ballots for late receipt | 23.1% | EAVS 2022 Comprehensive Report |
| States reporting drop boxes | 39 | EAVS 2022 Comprehensive Report |
| Drop boxes reported by remaining states | 12,917 | EAVS 2022 Comprehensive Report |
| Available on Election Day | 11,858 | EAVS 2022 Comprehensive Report |
| Available during early voting | 8,386 | EAVS 2022 Comprehensive Report |
What these statistics show
The federal and state numbers point in different directions, and that is the point
The federal prosecution data shows how rare election-fraud cases are relative to overall caseloads. The Michigan audit data shows how a state can review ballots at scale and still report no evidence of intentional misconduct or fraud by election officials. The EAVS data shows how administrative processes generate many other statistics around rejection, reporting, and infrastructure.
The combined picture is more informative than any single headline number:
- election-fraud matters are a tiny share of federal criminal activity in the dataset (GAO-19-485)
- case activity is geographically concentrated but still appears in most federal districts at least once (GAO-19-485)
- charge statutes are varied rather than singular (GAO-19-485)
- audit results can be large in scope without producing evidence of intentional misconduct (Michigan Post-Election Audit Report)
- election administration data includes a meaningful amount of non-fraud operational complexity, from rejection reasons to drop-box counts (EAVS 2022 Comprehensive Report)
A practical reading of the dataset
If you are using election fraud statistics to understand the scale of the issue, the safest interpretation is the most specific one the numbers support:
- federal election-fraud prosecutions are real but rare relative to overall caseloads (GAO-19-485)
- some districts carry a larger share of those cases than others (GAO-19-485)
- audits and administrative reports should be read as process data, not as automatic proof of fraud or absence of fraud (Michigan Post-Election Audit Report; EAVS 2022 Comprehensive Report)
- the best comparisons come from matching like with like: caseload to caseload, sample to sample, and administrative measure to administrative measure
A final data point to keep in view
One of the clearest lines in the full dataset is also one of the simplest: 49 of 94 federal judicial districts did not prosecute any election-fraud related cases from FY2001-FY2017 (GAO-19-485).
That single number captures the broader pattern well. The federal record is wide, but not dense. It is distributed, but concentrated. And the audit data in the dataset reinforces the same habit of careful reading: large-scale election review can be detailed, local, and methodical without implying a broad finding of intentional misconduct.