10 wrongful convictions in American history — and what freed them

10 wrongful convictions in American history, from false confessions to mistaken eyewitnesses, and the evidence that finally freed each man

A wrongful conviction rarely comes down to a single error. It usually takes a chain of them. An eyewitness picks the wrong face under pressure. A teenager signs a statement after hours of questioning. A forensic analyst gives testimony that sounds more exact than the science behind it. An appeals court declines to reopen a verdict it already approved. The 10 men described here served more than 190 years in prison combined for crimes they did not commit.

Their cases appear in the Innocence Project’s closed case records. The men are Antron McCray, Alejandro Hernandez, Bruce Godschalk, Byron Halsey, Alan Newton, Archie Williams, A.B. Butler, Calvin Willis, Alan Crotzer and Anthony Wright. They were convicted in New York, Illinois, Pennsylvania, New Jersey, Louisiana, Texas and Florida, and exonerated between 1995 and 2019. The shortest stretch behind bars was six years. The longest was 35.

What makes these wrongful convictions in American history worth studying now is the mechanics of the reversal as much as the injustice itself. Several men were cleared by DNA testing that did not exist when they were tried. One was freed by a fingerprint run through an FBI database decades after the print was collected. Another waited years while the evidence that could clear him was reported lost, then found in a police storage facility. In one case, the prosecutor’s office fought the retesting request that eventually proved the conviction wrong.

The causes repeat across the roster. False confessions were drawn from juveniles and vulnerable suspects. Eyewitnesses identified the wrong man with confidence and nothing to corroborate it. Informants had reasons to tell investigators what they wanted to hear. Serology testimony implied more certainty than the method could deliver. The fixes do not repeat. Each slide that follows covers one case, the evidence that sent the man to prison, and the specific break — scientific, legal or confessional — that finally undid the conviction.

Antron McCray was convicted at 15 on a confession with no physical evidence

Antron McCray was 15 when New York police obtained the statement that convicted him in the 1989 Central Park jogger case, and no physical evidence ever tied him to the attack. He served six years, and his conviction was vacated in December 2002 after Matias Reyes said he committed the assault alone and DNA testing confirmed it.

McCray was questioned for hours before giving investigators the account that anchored the case against him. He and four other teenagers — the group later known as the Central Park Five — went to trial on statements that conflicted with each other and with what investigators found at the scene. Jurors heard those confessions. They did not hear forensic proof, because none linked any of the five to the woman who was raped and beaten in the park.

The six years McCray served was the shortest term among the five. He and three co-defendants were prosecuted as juveniles. Korey Wise, who was 16 at the time of the attack, was tried as an adult and spent about 13 years in prison. Release settled nothing for McCray, who came home with a felony record for a crime he had not committed and kept saying so.

The reversal came from outside the original investigation. Reyes, by then serving a life sentence for a series of rapes and a murder, told authorities he had attacked the jogger by himself. DNA testing matched him to semen recovered from the scene — the same unidentified profile that had failed to match any of the five at trial, which prosecutors had attributed to an uncharged sixth attacker. A Manhattan judge vacated all five convictions in December 2002.

The case changed how courts and police departments weigh statements taken from minors, and it remains a central example in the argument that a teenager under sustained interrogation will say what ends the interrogation. New York now requires police to electronically record custodial questioning in homicide and other serious felony investigations.

Alejandro Hernandez was convicted on informant testimony and disputed statements

Alejandro Hernandez was convicted of the 1983 murder of 10-year-old Jeanine Nicarico on informant testimony and statements investigators attributed to him after prolonged questioning. No forensic evidence placed him at the scene. He spent 10 years in Illinois custody before prosecutors abandoned the case in 1995.

The prosecution of Hernandez and co-defendant Rolando Cruz in DuPage County rested on secondhand accounts — remarks the two men were said to have made, relayed in part by witnesses who stood to gain from cooperating with the state. The state sought the death penalty. Both men maintained they had nothing to do with the abduction and killing of the girl, who was taken from her home in Naperville.

The case started to come apart because of Brian Dugan. In custody for other killings, Dugan told authorities in 1985 that he had abducted and murdered Nicarico by himself, and his account included details about the crime. Prosecutors spent years discounting it. DNA testing applied to biological evidence from the case eventually matched Dugan and excluded Hernandez. Dugan pleaded guilty to the Nicarico murder in 2009.

An Illinois appellate court reversed Hernandez’s conviction, and the state chose not to retry him. The fallout did not stop with his release. A special grand jury indicted several law enforcement officials connected to the prosecution over their handling of evidence. All were acquitted at trial, but the indictments were rare enough to make the Nicarico investigation a fixed reference point in arguments about holding officials accountable for a failed case.

The case mattered in Illinois well beyond the two men it wrongly imprisoned. The Nicarico exonerations were among those that drove the state’s reckoning with capital punishment, which produced a moratorium on executions in 2000 and repeal of the death penalty in 2011.

What the prosecution showed is how far a case can travel on informant testimony and contested statements when no physical evidence supports either.

Bruce Godschalk’s exoneration came after prosecutors resisted DNA retesting for years

A confession and serology testimony kept Bruce Godschalk in a Pennsylvania prison for 15 years for rapes he did not commit. He went free in 2002 only after an independent laboratory, outside the prosecuting office’s control, ran DNA tests that excluded him.

Technology was not the holdup. By the time Godschalk began asking courts to order testing, DNA analysis could already settle who had left biological evidence at a crime scene. The district attorney’s office opposed his requests. The dispute dragged on for years while he stayed in a cell, long after that science had become routine in American courtrooms.

Testing was eventually carried out by a lab not answerable to the office that won the original conviction. The results ruled him out as the source of the evidence. His conviction was vacated and he walked out in 2002.

Prosecutors who secure a conviction have an institutional interest in its finality. Leaving the decision to retest with that same office creates a conflict of interest. Godschalk’s fight became a standard reference point in a specific argument: that access to post-conviction DNA testing belongs in statute, granted by a court under defined conditions, rather than left to a prosecutor’s discretion. All 50 states eventually adopted post-conviction DNA testing laws, though the conditions attached to them vary widely.

The forensic evidence against him is instructive on its own. Serology — the older practice of typing blood and other bodily fluids — could place a suspect within a broad class of possible contributors, a group that might include a large share of the population. It could not single out one person. Testimony about serological results often carried an air of precision the method never had, and juries in the years before DNA testing had little basis for discounting it.

Pair that kind of forensic testimony with a confession and the result was a conviction courts were slow to revisit, as Godschalk’s 15 years demonstrate.

Byron Halsey’s false confession unraveled after DNA implicated a different man

Byron Halsey was convicted in New Jersey of murder and sexual assault largely on the strength of a confession he gave to police. He served 19 years before DNA testing in 2006 excluded him and pointed to a different man, and his exoneration followed in 2007.

The statement attributed to Halsey became the spine of the prosecution’s case. No forensic result tied him to the killings. The physical evidence gathered at the scene was never analyzed in a way that could identify one individual, because that technology did not exist when he went to trial. Jurors heard a confession instead, and a confession tends to outweigh doubt about everything else in a case.

When the crime-scene material was finally run through modern DNA methods, the profile matched Clifton Hall, a neighbor of the victims who had never been charged in the case. The same testing ruled Halsey out entirely. Hall was later convicted of the murders, and Halsey walked out of prison after nearly two decades.

What the case demonstrates in plain terms is that a confession can sound convincing and still be false. The conditions that recur in false-confession cases include interrogations that stretch across many hours, investigators telling a suspect that evidence against him already exists, and vulnerabilities such as youth, limited literacy or cognitive impairment that make a suspect more willing to say whatever ends the questioning.

Halsey’s release fed arguments in New Jersey and elsewhere for recording interrogations in full rather than preserving only a final written statement composed after the session ended. A summary written by the officers who ran the questioning cannot show a reviewer how a suspect arrived at the words on the page, or which details he supplied and which were supplied to him. Without a recording, the gap between what was said in the room and what reached the jury cannot be audited at all.

Alan Newton’s evidence was declared lost for years before DNA cleared him

Alan Newton’s exoneration was blocked for years not by a witness or a confession but by a filing failure: the evidence he needed for DNA testing was reported missing, then found in police storage. Testing cleared him in 2006, after 21 years in New York prisons.

Newton asked for that testing repeatedly. Each time, the answer came back that the rape kit and related evidence from his case could not be located. A post-conviction DNA request is useless without biological material to analyze, so the reported loss worked as a closed door rather than a delay. He kept filing anyway.

The material had not been destroyed. It turned up in a city evidence facility, inside the same system that had reported it gone. Once it was produced and analyzed, the DNA excluded Newton as the source of biological material from the assault, and his conviction was vacated.

That sequence exposes a failure distinct from the interrogation-room and lineup errors behind most famous wrongful conviction cases. The science was available. The defendant was demanding it. The sample existed the entire time. What was missing was an evidence-management system able to find its own holdings on request.

Newton’s case became a reference point in arguments over evidence retention and cataloging — how long biological material from serious felonies should be kept, who is responsible for locating it, and whether a jurisdiction’s inability to find what it stores should carry legal consequences. A rape kit sitting unindexed in a warehouse is, in practice, the same as no rape kit at all.

His exoneration also marks a hard limit on forensic progress. Better testing corrects a wrongful conviction only when someone can retrieve the thing to be tested. Newton’s account of his own innocence never changed across those two decades; what changed was whether anyone could check it against the evidence the city already held.

Archie Williams was freed by a fingerprint database match, not DNA testing

A fingerprint, not a DNA sample, ended Archie Williams’s 35-year imprisonment in Louisiana. He was convicted of rape on the strength of an eyewitness identification and exonerated in 2019, after a print recovered from the crime scene was run through an FBI automated fingerprint database and matched a man who had never been a suspect.

The print had been in the case file since the original investigation. What did not exist at the time was any practical way to search it. Comparing a latent print against a large collection of known prints meant manual review by examiners, which confined searches to candidates investigators already had in mind. A print left by someone outside the suspect pool could sit in a file for decades without ever being identified.

That changed as the FBI’s automated fingerprint identification system came online and expanded nationally in the years after Williams’s trial, turning a slow manual comparison into a database query. When the print from his case was finally submitted for a search, it returned a name — and the name was not his.

The match did more than exclude Williams. It identified an alternative perpetrator, which replaces a conviction rather than simply casting doubt on it. The eyewitness identification that had carried the case for three and a half decades did not survive the result.

Williams served 35 years, the longest stretch among the cases collected here. His case shows that post-conviction innocence work is not limited to DNA. Fingerprints and other physical evidence collected before searchable databases existed can hold answers that nobody at the time had the tools to read.

The case also sets up a wider question for old files. Many wrongful conviction reviews start by asking whether biological evidence survived in testable condition. Williams’s exoneration points to a second line of inquiry: what else was collected at the scene, and what can current systems do with it that the original investigators could not.

A.B. Butler was cleared by DNA testing after an eyewitness misidentified him

DNA testing ended A.B. Butler’s rape conviction in Texas in 2000, after he had already served 16 and a half years. The case against him had rested on an eyewitness identification, and the genetic evidence excluded him as the source of the biological material recovered in the investigation.

Butler’s imprisonment began in the 1980s, before DNA profiling entered U.S. criminal courtrooms at all. A jury in that era weighing a sexual assault charge often had little to evaluate beyond the account of the person who survived the crime and the identification that followed. An identification delivered without hesitation can carry enormous weight with jurors, and nothing in the courtroom of that period was equipped to check it against the physical evidence sitting in an evidence locker.

By the time Butler’s case was reopened, that had changed. Genetic testing had matured into a method capable of matching or excluding an individual from a biological sample with a precision that no witness memory can approach. Applied to the evidence from his case, it excluded him, and Texas released him in 2000, after more than a decade and a half behind bars.

Mistaken identification is a recurring factor in convictions later undone by genetic testing, and Butler’s followed the familiar sequence: a witness account accepted as decisive, no forensic material measured against it, and a conviction that held until the sample was finally analyzed. Lineups and photo arrays of that period were typically assembled and administered by officers who knew which person was the suspect, which left no structural guard against steering a witness toward that answer.

His case is also a reminder of what the arithmetic of a delayed exoneration looks like in practice. Butler was not freed because a court found a procedural flaw in his trial, and not because a witness changed an account. He was freed because a technology that did not exist at his trial eventually reached evidence that had survived. The gap between the verdict and the test that undid it ran 16 and a half years.

Calvin Willis served 21 years after a mistaken eyewitness identification in a child rape case

Calvin Willis was convicted in Louisiana of raping a child, and he spent 21 years in prison before DNA testing of preserved evidence excluded him. He was exonerated in 2003.

The case against him turned on identification rather than physical proof. No forensic method available at the time of trial could independently confirm or contradict that identification, so jurors were left weighing a witness account against a denial.

Identification evidence involving a young victim carries particular difficulty. A child’s account of an assault is shaped by the questions asked afterward, by the way a photo array or lineup is presented, and by how many times the child is asked to look again at the same faces. Suggestion can enter at any of those points without anyone intending it. Once an identification is made and repeated, it tends to harden, and the witness’s confidence grows even when the underlying memory has not changed. A jury sees only a witness who sounds certain.

What eventually resolved the question was material from the original investigation that survived in testable condition for more than two decades. DNA testing performed on that evidence excluded Willis as the source of the biological material tied to the crime. The result did not simply weaken the identification. It replaced the only substantive evidence in the case with a scientific answer that pointed away from him, and his conviction was vacated in 2003.

His release followed the route that produced most exonerations of that period: old evidence, newer chemistry, and an answer different from the one a jury reached.

The case is often raised in arguments for mandatory retention of biological evidence in serious felony cases. Had the material from the original investigation been discarded at any point during those 21 years, nothing would have remained to test, and the identification that convicted him would have stood as the final word.

Alan Crotzer spent 24 years in prison before DNA testing proved his innocence

Alan Crotzer served 24 years in Florida prisons after eyewitness identification tied him to a rape and robbery he did not commit. DNA testing on evidence retained from the original case excluded him, and he was exonerated in 2006.

Crotzer’s sentence ranks among the longest on this list, and the evidence behind it was the kind juries find hardest to doubt. An identification delivered in court by someone who survived the crime carries a weight that cross-examination rarely undoes. When he was tried, no genetic test existed to measure that identification against biological material recovered from the scene.

Mistaken identification appears in more DNA exoneration cases than any other single factor, and it recurs for reasons that have nothing to do with a witness lying. Memory is reconstructive rather than fixed. A witness assembles a face out of fragments, and each later viewing — a photo array, a lineup, the defendant seated alone at a table in court — can reshape what gets recalled. Certainty also hardens over time, so the confident testimony a jury hears years after a crime may bear little resemblance to how sure the same person was when police first asked.

His release turned on an administrative detail rather than a legal argument. Evidence from the original investigation still existed, and it was intact enough to analyze. Had it been destroyed or discarded in the intervening decades, as happens with plenty of older case files, nothing would have been left to test. The identification would have stood. Preservation rules vary by state and by county, and biological material from a decades-old file survives as often by accident as by policy.

Crotzer left prison in 2006 having lost the years in which most people build a work history, buy property and pay into Social Security. DNA testing capable of producing that result had been in use for years before anyone applied it to his case.

Anthony Wright was convicted as a teenager and exonerated decades later by DNA

Anthony Wright served 25 years in Pennsylvania prisons for a rape and murder he did not commit. He was convicted as a teenager and cleared in 2016, when DNA testing of crime-scene evidence excluded him and matched another man. The genetic exclusion alone did not end the case — prosecutors took him to trial a second time.

That second trial is what separates Wright’s case from most DNA exonerations. A laboratory result identifying a different source of biological evidence does not automatically vacate a verdict or close a prosecution. The state retains the option of retrying a defendant on its original theory of the crime, and in Wright’s case it did exactly that, more than two decades after the first conviction. A jury heard the case again and acquitted him, which is what finally ended his incarceration.

His age at conviction compounded the loss. He was young when he entered the system, and the years he lost covered the span in which most people build careers, raise families and accumulate the ordinary records of a life outside prison.

The evidence that cleared him sat in storage the entire time, preserved from the original investigation and available for analysis once testing methods advanced far enough to produce a usable profile. The gap between when that testing became technically possible and when it was actually performed accounts for a meaningful share of the years Wright spent behind bars.

His 2016 exoneration is the most recent of the 10 cases covered here, and it corrects a common assumption about how these reversals work. A DNA match to someone else is not an instant release order. It is a starting point. Undoing a conviction still requires courts to act, prosecutors to concede or lose, and in Wright’s case, a second jury to say what the first one got wrong.