Wrongful Conviction Remedies: Reopening a Final Verdict Is Only Half the Fight

The United States never built a national commission. It layers a narrow federal habeas remedy, 28 U.S.C. Sections 2254 and 2255, hemmed in by the one-year clock the Antiterrorism and Effective Death Penalty Act of 1996 (AEDPA) imposes, on top of a bottom-up innocence movement that has produced more than 4,000 documented exonerations and over 32,000 years of wrongful imprisonment since 1989. DNA evidence, despite driving the public narrative around wrongful conviction, accounts for only about 15 percent of it. Five causes recur across nearly every system studied instead: eyewitness misidentification, false confessions, forensic science error, official misconduct, and ineffective assistance of counsel.

The United Kingdom shows the compensation divide most starkly: its test requires proof "beyond reasonable doubt" of innocence, a standard the European Court of Human Rights' Grand Chamber upheld in 2024 even though both men involved had their convictions quashed as unsafe. Italy runs two compensation tracks nearly 150 times apart in volume, wrongful pretrial detention against true judicial error, and Germany's own reopening statute was amended again as recently as December 2025. None of this is settled law anywhere studied here.

Post-Conviction Review Mechanisms: A Commission Model and Four Ordinary-Court Models

Every system compared here built a way to reopen a conviction once final, but the five jurisdictions split into two structural families. England, Wales, and Northern Ireland, together with Scotland, built independent commissions with statutory investigative powers standing between the convicted person and the appeal court. The United States, France, Germany, and Italy route the same function through ordinary courts instead, with no specialist body in between.

The United States relies on federal habeas corpus, reshaped by AEDPA's one-year filing clock and its bar on relief for any claim a state court already decided, unless that state court's decision was contrary to, or an unreasonable application of, clearly established federal law. AEDPA turns federal habeas into a deferential check on the state court's own process rather than a fresh look at guilt or innocence.

The Criminal Cases Review Commission (CCRC), covering England, Wales, and Northern Ireland, was created by the Criminal Appeal Act 1995, a direct response to the Birmingham Six miscarriage of justice, and began work on 31 March 1997. It applies Lord Bingham's real possibility test: more than an outside chance, less than a probability.

The Commission has referred more than 870 cases since 1997, averaging roughly 33 a year; in 2024-25 it referred 31 cases, up 24 percent year over year, and 18 of the 27 heard appeals were allowed, a success rate near two-thirds. That test now sits under direct reform pressure: the Westminster Commission on Miscarriages of Justice concluded in its March 2025 report that the CCRC is underfunded, insufficiently independent, and applies a merits bar set too high, and recommended a Law Commission review.

Scotland runs a separate commission, the Scottish Criminal Cases Review Commission (SCCRC), established under devolved Scottish legislation in April 1999. Between 1999 and 2014 it reviewed more than 1,800 cases and referred 122, of which 70 succeeded on appeal, a rate close to 67 percent, within a few points of the CCRC's own 2024-25 figure. That similarity is a comparative data point, not proof of anything more, reflecting this article's own reading of two independently reported statistics rather than any claim either commission has made.

France reopens a final conviction through revision, governed by Articles 622-626 of the Code de procedure penale, available where a new fact emerges "likely to establish the innocence of the convicted person or to raise doubt about their guilt." The Loi du 20 juin 2014 merged what had been two separate procedures, revision for new facts and reexamen following an adverse ruling from the European Court of Human Rights, into a single cour de revision et de reexamen.

Germany's Wiederaufnahme, governed by Section 359 of the Strafprozessordnung (StPO), lists the statutory grounds for reopening a case in the convicted person's favor: forged evidence, false testimony, judicial misconduct, or new facts capable of a milder outcome. Section 362 No. 5 StPO, effective 30 December 2021, allows for the first time a retrial against a previously acquitted person for murder or crimes against humanity on new "clear proof of guilt," a reversal of direction from every other favor-of-the-convicted mechanism here, and one under active constitutional challenge as of 2024 academic commentary.

Italy's revisione, governed by Articles 629-647 of the Codice di procedura penale, is available at any time, even after a sentence has been fully served, and is explicitly favor rei: revision against an acquittal is never permitted. The convicted person, close relatives, and heirs may request it, and so may the procuratore generale (prosecutor general), a prosecutorial self-correction channel with no true equivalent in the CCRC or SCCRC commission model.

The same civil-law-versus-common-law structural divide reappears in how these same systems route negligence liability, the subject of the companion article on comparative negligence.

Why Convictions Go Wrong: Five Recurring Causes Across Every System Studied

Five causal categories recur across nearly every system studied, though the exact percentages vary by dataset, time period, and methodology.

Eyewitness misidentification is the most cited cause in DNA-specific data: the Innocence Project puts it at 69 percent of DNA exonerations (252 of 367 cases), against the National Registry of Exonerations' broader figure of 43 percent across all exoneration types. The two percentages measure different bases and should not be merged.

False confessions appear in 24 percent of roughly 289 DNA-based reversals per the Innocence Project, against 13 percent of all documented wrongful convictions in the National Registry's broader dataset. The average interrogation preceding a false confession runs 16 hours; 63 percent of false confessors were under 25, and 32 percent under 18.

Forensic science error has a single, stark anchor: the FBI's own internal review found testimony errors in more than 90 percent of reviewed microscopic hair-comparison cases, with 26 of 28 examiners overstating matches in more than 95 percent of the 268 trials reviewed, prompting notice to defendants and prosecutors in 46 states plus D.C. Pattern-matching forensic disciplines more broadly are implicated in more than one-quarter of the 329 DNA-exoneration cases studied.

Official, prosecutorial, and police misconduct shows up across a wide range depending on dataset and year: 71 percent of fully overturned convictions in 2024, and 59 percent of all National Registry-documented wrongful convictions historically. Common forms include suppression of exculpatory evidence, witness tampering, and coercive interrogation tactics.

Ineffective assistance of counsel is governed by Strickland v. Washington, 466 U.S. 668 (1984): a two-prong test requiring objectively unreasonable performance plus a reasonable probability of a different outcome. Only about 8 percent of state habeas ineffective-assistance claims succeed, making it the most-raised post-conviction claim and one of the hardest to win.

A sixth pathway recurs alongside these five: false guilty pleas, driven in the United States by the trial penalty, the gap between a pre-trial plea offer and the sentence imposed after conviction at trial. The companion article on plea bargaining's international adoption and resistance documents the trial penalty and the share of exonerations it accounts for; this article names the pathway rather than restating those figures.

DNA's Real Share: The Innocence Project, the National Registry, and What DNA Actually Explains

Are most wrongful convictions overturned using DNA evidence? No. DNA exonerations make up only about 15 percent of all documented US exonerations; the large majority involve other evidence, including recanted testimony, uncovered official misconduct, or non-DNA forensic re-analysis.

The Innocence Project, founded in 1992, had participated in 254 DNA-based exonerations as of April 2026. Its exonerees served an average of 14 years; 10 percent served 25 years or more.

The National Registry of Exonerations, run by the Newkirk Center at the University of Michigan and Michigan State University, is the broader field-wide tracker. As of June 2025 it listed 3,698 exonerations since 1989, plus 463 pre-1989 cases and 46 "Groups Registry" mass-exoneration cases tied to a single government actor's misconduct: more than 4,000 exonerations in total, and over 32,000 years of wrongful imprisonment. Its 2023 annual report recorded 153 exonerations, nearly 84 percent involving people of color.

Conviction Integrity Units, prosecutor-office divisions that review the office's own past convictions, are a novel non-judicial channel with no equivalent in the models described above. They have doubled over the last ten years but remain concentrated in large urban jurisdictions; no comprehensive nationwide count of active units exists, so the growth is a documented trend, not a current total.

Post-conviction DNA testing statutes now exist in all 50 US states, complete as of Oklahoma's adoption on 24 May 2013, a decade after the 2004 federal Innocence Protection Act supplied the model most states followed.

Two Separate Fights: Exoneration and Compensation Across Five Systems

Section 133 of the Criminal Justice Act 1988 is the UK's compensation provision, and its amendment history is where sources most often go wrong. It was amended by section 175 of the Anti-social Behaviour, Crime and Policing Act 2014, which inserted a statutory definition requiring a new or newly discovered fact to show "beyond reasonable doubt" that the applicant did not commit the offence.

That definition is the product of a three-stage doctrinal arc: R (Mullen) v. Secretary of State for the Home Department [2004] UKHL 18 read section 133 narrowly, covering only someone later shown demonstrably innocent. R (Adams) v. Secretary of State for Justice [2011] UKSC 18 broadened the test to cover a conviction "so undermined that no conviction could possibly be based" on the remaining evidence. The 2014 Act then legislated Adams back down toward Mullen.

Compensation caps stand at 1 million pounds for 10-plus years' imprisonment and 500,000 pounds for under 10 years, unchanged since 2008. A Statutory Instrument, SI 2025/1139, raises both caps by 30 percent, to 1.3 million pounds and 650,000 pounds; its precise in-force date was not independently confirmed in the research behind this article and should be checked against legislation.gov.uk before it is treated as settled.

Nealon and Hallam v. the United Kingdom, decided by the European Court of Human Rights' Grand Chamber in 2024, is the doctrinal capstone of that arc. The Grand Chamber held that the narrow 2014 test does not violate the presumption of innocence guaranteed by Article 6(2) of the European Convention on Human Rights, because a state may legitimately limit which quashed convictions qualify for compensation, so long as the refusal itself does not impute guilt, even though both men's convictions had been quashed as unsafe.

Italy runs two structurally separate compensation tracks with a wide volume gap between them. Ingiusta detenzione, wrongful pretrial detention, produced 32,262 cases from 1992 through October 2025, nearly 949 a year, with more than 925 million euros paid. Errori giudiziari, true judicial error reached through revisione, produced just 222 cases from 1991 through 2022, 7 a year, with 76.2 million euros paid, a gap of nearly 150 to 1 against the pretrial-detention track. That gap suggests Italy catches most errors at the pretrial-custody stage, before conviction becomes final, unlike the US model's near-exclusive post-conviction focus.

The United States runs a 38-state-plus-District-of-Columbia patchwork of compensation statutes, with the federal rate at 50,000 dollars per year and state amounts ranging from 5,000 dollars a year, capped at a 25,000-dollar total, in Wisconsin, to 200,000 dollars a year in Washington, D.C. States add their own exclusions on top: a prior unrelated violent-felony record, or having entered a guilty plea rather than been convicted at trial, can disqualify an otherwise-exonerated person from compensation entirely.

Article 3 of Protocol No. 7 to the European Convention on Human Rights, in force since 1988, obligates compensation for a reversed final conviction but leaves the standard to domestic law, which is why the UK, Italy, and US standards diverge so sharply. One legal-academic source characterizes the provision as successfully invoked in only one case decided by the European Court of Human Rights directly; that figure comes from a single source, not an independently confirmed search of the Court's own case-law database. The practical compensation-forcing work in this area runs mostly through Article 6(2) litigated against domestic schemes, as Nealon and Hallam illustrates, rather than through Protocol 7 Article 3 on its own.

This is a different compensation universe from the state funds that compensate crime victims generally: there, the state pays for someone else's crime; here, it pays for its own error in convicting the wrong person. The companion article on victim compensation systems covers that parallel system.

Five Cases That Show What the Remedy Does and Does Not Fix

Victor Nealon was convicted in 1997 of attempted rape and served 17 years and 3 months before DNA on the victim's clothing pointed to an unknown male. Sam Hallam was convicted in 2004 of murder and served 7 years and 7 months on identification evidence later undermined. Both men had their convictions quashed as unsafe. Both were refused compensation under the 2014 test described above.

The Birmingham Six had their convictions quashed by the Court of Appeal on 14 March 1991, the miscarriage of justice behind the CCRC's creation. Compensation followed a decade later, in 2001, ranging from 840,000 to 1.2 million pounds per person.

The Central Park Five had their convictions vacated in 2002. New York City settled with the five men in 2014 for 41 million dollars total, roughly 7.1 million each for four of them and 12.2 million for the fifth, Korey Wise, who had served six additional years; the city admitted no wrongdoing.

The Outreau affair began in November 2001, when 18 people were held in pretrial detention for as long as three years without material evidence, driven largely by one investigating judge's intuition and one co-defendant's false testimony. Ten of the accused adults were convicted at trial before the Cour d'assises in Saint-Omer between May and July 2004; on appeal to the Paris Court of Appeal in November 2005, six of those ten were acquitted. The acquittals produced a 40 percent drop in child sexual assault convictions over the following decade, a long-tail cost attributed to prosecutorial and jury caution.

Amanda Knox was acquitted definitively of murder by the Corte di Cassazione in 2015, a complete remedy on the central charge. A separate slander conviction, for a false accusation she made during a night of police questioning, was never overturned, and received final confirmation from the Corte di Cassazione on 23-24 January 2025.

These five cases trace the outer edges of what a remedy actually does. Birmingham Six and Central Park Five show compensation eventually arriving, at very different scales. Nealon and Hallam show it can be refused entirely even after a quashed conviction. Knox shows exoneration itself can be partial.

Fixing the Front End: Recording, Lineups, and Forensic Oversight

Reopening a wrongful conviction and compensating for it are back-end remedies, arriving only after the harm is done. The reforms here target the front end instead, reducing how often a wrongful conviction happens at all.

More than half of US states plus D.C. now require recording of certain custodial interrogations, by statute or court decision. New Hampshire's SB 261 and Massachusetts's "An Act Preventing False Confessions" (H1847/S1136, introduced February 2025) are active efforts to expand that list further.

Double-blind, sequential lineup administration, described by researchers as the single most important reform for eyewitness evidence integrity, is now required by legislation or rule in 19 states. Indiana's 2025 law was the first to pair an eyewitness-identification reform with facial-recognition-technology guidelines for police, a pairing no other state has yet matched.

The 2009 National Academy of Sciences report, "Strengthening Forensic Science in the United States: A Path Forward," found forensic science broadly lacking rigorous scientific grounding and called for a fully independent oversight body removed from law enforcement and prosecutorial control. That recommendation was never implemented. The Department of Justice and the National Institute of Standards and Technology created the National Commission on Forensic Science in 2013 as a compromise body instead, only a partial response.

What Changed Between 2020 and 2026

The United Kingdom saw the most concentrated pressure on its review system in this window. CCRC Chair Helen Pitcher resigned in February 2025 amid the Westminster Commission's critique described above; commentary treats the resignation as secondary to the deeper structural critique of the real possibility test.

Germany's Section 359 StPO was amended again on 22 December 2025 (BGBl. 2025 I Nr. 349). The substance of that amendment was not independently confirmed in the research behind this article; what can be said is that it occurred, the latest entry in continued legislative engagement with the Wiederaufnahme mechanism, alongside the more controversial Section 362 No. 5 StPO reform.

The United States produced two concrete state-level developments. Georgia's Wrongful Conviction and Incarceration Compensation Act, signed 14 May 2025, sets compensation at 75,000 dollars a year, 25,000 dollars more if the person was wrongly sentenced to death, inflation-adjusted starting 1 January 2026, confirmed signed law. Illinois's 2026 bill, raising its compensation cap to 50,000 dollars a year and extending eligibility to children for the first time, passed the General Assembly and was sent to the Governor; its final signature status was not confirmed as of this writing and should not be treated as settled.

Across all five jurisdictions studied, 2024 through 2026 shows active legislative and judicial attention to both reopening mechanisms and compensation standards. This is not a settled area of law.

Frequently Asked Questions About Wrongful Conviction Remedies

Does overturning a wrongful conviction automatically bring compensation with it?

No. Every system compared here treats winning a reversal and winning compensation as separate legal fights, often decided under considerably tougher standards, as the Nealon and Hallam case above illustrates.

Is DNA evidence behind most wrongful conviction reversals in the United States?

No. DNA-based exonerations account for roughly 15 percent of the documented total; the rest are corrected through recanted testimony, exposed official misconduct, and non-DNA forensic review.

Does international human rights law force every country to pay for a reversed conviction?

Not in any strong sense. Protocol No. 7, Article 3 to the European Convention on Human Rights requires compensation in principle but hands the actual standard to domestic law, which is why a state can lawfully narrow who qualifies, provided the narrowing itself does not imply guilt.

Where the Rest of Comparative Criminal Justice Lives

This article closes out Comparative Criminal Justice by asking what happens once a system's guilt-finding process turns out to have been wrong. The other four articles in this series cover how guilt is established, negotiated, and repaired in the first place. Victim compensation systems covers the parallel system that pays crime victims generally, discussed above as a contrast to wrongful-conviction compensation. Comparative negligence traces the same commission-versus-ordinary-court structural divide covered here, applied instead to negligence liability. Restorative justice models worldwide examines a genuinely separate question, since restorative justice presumes guilt was rightly established in the first place. Plea bargaining's international adoption and resistance covers the trial-penalty pressure behind the false-guilty-plea pathway named earlier in this article.