Wrongful Conviction Research & Exoneration Data

The National Registry of Exonerations — a joint project of the University of California Irvine, the University of Michigan Law School, and Michigan State University College of Law — is the most comprehensive, continuously updated database of documented U.S. exonerations, tracking each case's contributing factors, demographics, and outcome. The Innocence Project maintains a separate, related dataset focused specifically on its own client cases (primarily DNA exonerations).

These datasets are the empirical backbone for nearly everything Xonorate and other organizations say about how often, and in what ways, the criminal legal system convicts innocent people — and understanding their scope and limits matters for using their numbers responsibly.

Wrongful convictions are, by definition, difficult to count — a case only becomes a documented "exoneration" after a lengthy, uncertain process of discovery, litigation, and formal reversal. That means the tracked numbers are a real but almost certainly incomplete picture: they reflect confirmed, documented cases, not the full universe of wrongful convictions that may never be identified or overturned.

The research also matters because it reveals recurring, structural patterns rather than isolated errors — the same contributing factors (misidentification, informant testimony, forensic error, official misconduct, inadequate defense, false confessions) appear again and again across thousands of cases, which is what makes systemic reform, not just individual case correction, a coherent policy response.

Credible wrongful-conviction data comes from a small number of rigorous sources, each with a distinct scope:

• The National Registry of Exonerations tracks documented exonerations across the full range of crime types, not just those involving DNA, and publishes both a continuously updated live database and dated annual reports.
• The Innocence Project's own dataset tracks its client cases specifically — a smaller, DNA-exoneration-heavy population that is not representative of the wrongful-conviction problem as a whole, but offers detailed case-level data on its own cases.
• Independent academic studies occasionally attempt to estimate the overall rate of wrongful conviction within a defined population — for example, a rigorous 2014 statistical study estimated a minimum 4.1% false-conviction rate among people sentenced to death, and a 2017 NIJ-funded study estimated an 11.6% wrongful-conviction rate specifically among a sample of pre-DNA-era Virginia murder and sexual-assault convictions. Both are important because they attempt to measure a rate within a defined population, rather than simply counting known cases — but neither should be generalized beyond the specific population each study actually examined.

Live counters and dated annual reports serve different purposes. The Registry's live "contributing factor" statistics update continuously as new exonerations are logged, while its annual reports give a fixed, citable snapshot for a specific year — useful for year-over-year trend claims. Any statistic drawn from the live counter should be paired with the date it was pulled, since the numbers shift.

Documented contributing factors, current as of the most recent pull from the Registry's live data: perjury or false accusation (64%), official misconduct (61%), false or misleading forensic evidence (29%), mistaken witness identification (27%), and false confession (13%) — cases frequently involve more than one factor, so these figures do not sum to 100%.

Demographic patterns are well documented. Recent annual Registry data shows a substantial majority of exonerees are people of color, and Black exonerees are disproportionately represented relative to the general population — a pattern researchers and the Registry itself connect to the disproportionate rate of official misconduct documented in cases involving Black defendants specifically.

"Professional exonerators" — not just individual litigation — now drive most outcomes. The Registry's most recent annual reporting found that dedicated innocence organizations and prosecutor-run conviction integrity units, often working together, were involved in the clear majority of a recent year's exonerations.

Compensation remains uneven. Federal law (28 U.S.C. § 2513) provides compensation for federal wrongful convictions, capped by statute, but this applies only to federal cases; a substantial number of states have no wrongful-conviction compensation statute at all, and even where compensation exists, research documents significant barriers to actually obtaining it, alongside separate, serious challenges around mental health and reentry after release.

When using or citing wrongful-conviction data, documented practices worth following:

• Always identify which dataset a statistic comes from — the full National Registry database, the Innocence Project's own client cases, or an independent academic study — since they measure different, non-interchangeable populations.
• Pair any live-counter statistic with the date it was pulled, since the Registry's running figures change as new exonerations are added.
• Don't generalize a study's rate estimate beyond its actual scope — a 4.1% capital-case estimate or an 11.6% estimate from a specific Virginia sample should never be quoted as a general wrongful-conviction rate for the criminal justice system as a whole.
• Remember that documented exonerations are a floor, not a ceiling — they reflect confirmed, discovered cases, not the full scope of wrongful convictions, many of which likely go undiscovered.
• Look for corroboration across sources where possible, since even the best datasets rely on case-by-case researcher classification of "contributing factors," which involves judgment calls.