Skip to content
Search resources Open menu

False Confessions

False Guilty Pleas

Published

Snap Shot

  • The question: Why do some innocent defendants plead guilty, how does a prior confession affect plea decisions, and what safeguards does the law provide?
  • What the research shows: In laboratory studies using an academic-misconduct paradigm, guilty participants accepted plea offers far more often than innocent ones, but a sizable share of innocent participants also accepted (about 40% to more than half in some conditions, and more when advised to plead guilty). Large sentence differences, pretrial detention, and a confession in the case increased willingness to plead, and about a quarter of the exonerations in the national registry followed guilty pleas.
  • What it does not show: No study measures how often guilty pleas in general are false. The laboratory studies use students and minor stakes, the survey data rely on self-report, and the exoneration data describe known errors, not the large majority of pleas entered by people who committed the offense.
  • The law: A guilty plea must be knowing, intelligent, and voluntary and must have a factual basis (Boykin; Brady v. United States; Fed. R. Crim. P. 11; R.C.M. 910; Ariz. R. Crim. P. 17). A plea entered on competent advice generally waives claims that an earlier confession was coerced (McMann; Tollett; Mil. R. Evid. 304(f)(8)). Pleas entered while maintaining innocence may be accepted in federal court at the court’s discretion (Alford) but are not among the pleas R.C.M. 910 allows.
  • Issues for each side: For prosecutors and judges, the research bears on the depth of the factual-basis inquiry and on disclosure before the plea; for defense counsel, on assessing the reliability of any confession before advising a plea. The research identifies conditions that raise risk; it does not show that any particular plea is false.

Most criminal convictions in the United States are obtained by guilty plea rather than trial, and the Supreme Court has described the system as one of pleas rather than trials (Lafler v. Cooper, 2012). A guilty plea is more than an admission; the Supreme Court has described it as itself a conviction (Boykin v. Alabama, 1969). It is made in court, often under oath, and with the advice of counsel, and most pleas are presumed to be entered by people who committed the offense, although the proportion that are false is unknown. Research and exoneration data show, however, that some innocent people plead guilty, and that a prior confession to police is associated with a higher likelihood of a later plea.

The question reaches court in several ways: when a defendant seeks to withdraw a plea, when a military judge conducts the providence inquiry, when post-conviction relief is sought on the ground of innocence, and when defense counsel must decide what to advise a client who has given a disputed confession. It also bears on how far a guilty plea can be relied on as corroboration in related proceedings.

This paper summarizes the research on true and false guilty pleas, the relation between confessions and pleas, and the federal, military, and Arizona law that governs the acceptance and withdrawal of pleas. Confessions themselves are discussed in the other papers in this hub; vulnerabilities that affect both confessions and pleas are summarized in “Risk Factors for False Confessions: An Overview,” and plea issues for autistic defendants are addressed in “Charging, Diversion, and Plea Options in Autism Cases” in the Autism Online Offense resource.

Pleas in Known Wrongful Convictions

The National Registry of Exonerations reported that 879 of its first 3,618 exonerations, about 24%, involved convictions by guilty plea (as reported in Kassin et al., 2025). In a survey of defense attorneys, 89% reported having represented clients who pleaded guilty while maintaining their innocence, and 45% reported clients who pleaded guilty whom they believed to be innocent (Helm et al., 2018, as cited in Kassin et al., 2025). In a self-report survey of 1,249 people with mental disorders in the justice system, reports of having falsely pleaded guilty ranged from 27% to 41% across sites, and false admissions to murder and rape were rarely reported (Redlich et al., 2010).

These figures measure different things and should not be combined. Exoneration counts describe known wrongful convictions that could be proven, and the plea share among them is not a rate of false pleas among all pleas. Undoing a conviction by plea is also difficult, in part because a plea often waives the right to appeal (Kassin et al., 2025), so such cases may receive less later review than trial convictions, and the plea share among exonerations may understate or overstate the share among all wrongful convictions. Attorney surveys report impressions, not verified innocence, and self-reports of false pleas cannot be checked and may include people who believed they had a defense rather than people who did not commit the act. Together, they establish that false guilty pleas occur; they do not establish how often.

Laboratory Studies of Plea Decisions

Experimental research has adapted the cheating paradigm from confession research, in which students are accused of an academic violation after having cheated or not, and are offered a lenient resolution if they admit it or a harsher one if a hearing finds them responsible. In one small, low-stakes laboratory study, more than half of innocent participants accepted an offer to admit guilt in exchange for a benefit (Dervan & Edkins, 2013); the stakes and time frame differed from those of real cases. In a study of 328 students, the false plea rate was 40.7% and the true plea rate among guilty participants was 73.8%; the false and true confession rates in parallel conditions were 32.5% and 88.4%, and innocent participants gave different reasons for refusing a plea than for refusing a confession (Wilford & Wells, 2018). Guilty participants were consistently more likely to accept than innocent ones.

Advice matters. Innocent students advised by an advocate to go to a hearing falsely pleaded guilty 4% of the time, compared with 35% without an advocate, 47% when given educational information, and 58% when advised to plead guilty; advice affected innocent participants more than guilty ones (Henderson & Levett, 2018). Other studies summarized in the 2025 scientific review report that large sentence differences, an anchoring effect from a high threatened sentence, offers that avoid incarceration altogether, and release from pretrial detention increased innocent participants’ willingness to accept (Kassin et al., 2025). These are useful findings about decision processes, but the stakes in the laboratory are small and the decision is made in minutes, without the investigation, counsel, and time that precede a real plea.

How a Confession Leads to a Plea

A confession affects plea decisions mainly through its effect on the expected outcome at trial. Defendants and their lawyers who believe a confession will lead to conviction have reason to accept an offer, and the research is consistent with that account. In a hypothetical case, defense attorneys recommended plea acceptance more often when a confession was present (86.9%) than when it was absent (77.3%), and defense attorneys in another study recommended a plea in 3% of weak-evidence cases without a confession, 10% when the client had confessed but the confession was ruled inadmissible, and 39% when it was admissible (Bushway et al., 2014, as cited in Kassin et al., 2025; Hellgren & Kassin, 2022). In a simulation in which participants played a defendant, a confession produced more plea acceptances than eyewitness evidence, for guilty and innocent participants alike (DiFava et al., 2024).

The same pattern appears in actual cases. In a sample of presumably guilty defendants, about 70% of those who denied involvement to police later pleaded guilty, while those who gave partial or full confessions pleaded guilty at rates of 97% to 100% (Redlich et al., 2018). In an analysis of registry data comparing exonerations after trial with exonerations after plea, a false confession was the only canonical contributing factor that increased the likelihood of a false guilty plea, with an odds ratio of 4.28 (Redlich et al., 2025); both groups consist of cases already shown to be wrongful convictions, so the figure does not show how often confessions lead to false pleas generally. Among DNA exonerations, a false confession by the defendant or a co-defendant was present in 60% of false-plea cases compared with 24% of trial cases (Cooper et al., 2019, as reported in Kassin et al., 2025; the number of cases behind each percentage was not available to us, and the figures describe known DNA exonerations, not how common false pleas are). These data support the view that a confession, true or false, makes a guilty plea more likely; they also reflect the reasonable judgment that a confession is strong evidence, and the registry analyses describe only cases already shown to be wrongful convictions.

Pleas, Memory, and the Factual Basis

A defendant’s admission of the facts during the plea colloquy is often treated as confirming guilt. Research on plea paperwork suggests caution about how much a written form adds: an analysis of 208 tender-of-plea forms found that their content varied widely and that most exceeded the reading level of most defendants, and the authors concluded that forms cannot replace a full colloquy and meaningful attorney-client discussion (Redlich & Bonventre, 2015). Judgments of voluntariness may also depend on vantage point. In two online studies (N = 202 and N = 626) in which participants watched materials about a defendant deciding how to plead, those asked to take the defendant’s role rated the plea as more voluntary than observers did, contrary to the authors’ prediction; pleas were rated more voluntary when the defendant was guilty and when the defendant had a voice in the decision (Fessinger & Kovera, 2022). Some defendants plead to facts they do not remember, for example after intoxication; military rules permit an accused who cannot recall events to plead on the basis of evidence he believes to be true (R.C.M. 910(e), Discussion). Such a plea may be entirely accurate, but its factual basis then rests on the other evidence rather than on the accused’s own recollection. Memory loss and intoxication are discussed in the Alcohol Blackouts and Memory in the Courtroom resources on this site.

Limits of the Evidence

The experimental studies use college students accused of academic violations, with consequences that are small and immediate, and they compress into minutes a decision that ordinarily takes weeks or months with counsel. Their strength is that guilt is known. The field and registry studies concern real cases, but exonerations after pleas are a small and selected group, the plea share among them is not a prevalence rate, and the survey studies rely on unverified self-report. Most of the recent findings on confessions and pleas are summarized here as reported in the 2025 scientific review, and some use hypothetical decisions by attorneys or participants. Few of these studies analyzed sex or gender: Wilford and Wells (2018), whose sample was 55% female, did not report sex differences; Fessinger and Kovera (2022) used samples roughly evenly divided between men and women and did not report sex differences; and the other studies reviewed here did not report them. Research identifies conditions under which innocent people are more likely to plead guilty; it cannot establish whether a particular plea was false.

Questions to Ask About This Research

When the laboratory plea rates (Dervan & Edkins, 2013; Wilford & Wells, 2018; Henderson & Levett, 2018) are cited, ask what the participants stood to lose, how long they had to decide, and whether they had counsel; also ask about the guilty participants’ much higher plea rates, which show that pleas track guilt in the same studies. When the registry figure that about a quarter of exonerations followed pleas is cited, ask whether it describes a rate of false pleas or a share of known errors. When the confession-to-plea findings (Redlich et al., 2018, 2025; DiFava et al., 2024) are cited, ask whether the confession in this case was examined for reliability before the plea, and what independent evidence supported the plea. These questions are explained in “Reading Research Claims: A Guide for Attorneys” on this site.

Guilty Pleas and the Law

Federal and Military Standards

A guilty plea waives the privilege against self-incrimination, the right to a jury trial, and the right to confront witnesses, and the record must show that it was knowing and voluntary (Boykin v. Alabama, 1969). In Brady v. United States (1970), the Court upheld a plea entered to avoid a possible death sentence, while stating that it would have serious doubts if offers of leniency substantially increased the likelihood that defendants advised by competent counsel would falsely condemn themselves. Prosecutors may threaten more serious charges if a defendant declines to plead (Bordenkircher v. Hayes, 1978), and defendants are entitled to effective assistance of counsel in plea bargaining (Missouri v. Frye, 2012; Lafler v. Cooper, 2012). The Constitution does not require disclosure of impeachment information before a plea (United States v. Ruiz, 2002). Under Federal Rule of Criminal Procedure 11, the court must address the defendant personally, determine that the plea is voluntary, and find a factual basis.

Case Example. In North Carolina v. Alford (1970), the defendant was indicted for first-degree murder, a capital charge. Facing strong evidence, his attorney recommended a plea, and the prosecutor accepted a plea of guilty to second-degree murder. The trial court heard testimony from witnesses before accepting it. The defendant pleaded guilty while disclaiming guilt, saying he did so because of the threat of the death penalty, and was sentenced to 30 years. The Supreme Court held that accepting the plea was not constitutional error, because he had intelligently concluded that a plea served his interests and the record contained strong evidence of actual guilt. The case shows that the law permits a plea without an admission, and that the safeguard it relies on is the independent evidence in the record, not the defendant’s words.

A plea has particular consequences for a disputed confession. A defendant who pleaded guilty on competent advice generally may not later attack the plea on the ground that it was motivated by a confession he claims was coerced (McMann v. Richardson, 1970), and a guilty plea generally forecloses independent claims of earlier constitutional violations, leaving challenges to the voluntariness of the plea and the adequacy of counsel’s advice (Tollett v. Henderson, 1973). A challenge to the confession therefore ordinarily has to be resolved before the plea, or preserved through a conditional plea, which in federal court requires the consent of the court and the government (Fed. R. Crim. P. 11(a)(2)). Before sentencing, a federal defendant may withdraw an accepted plea for a fair and just reason; after sentencing, the plea may be set aside only on direct appeal or collateral attack (Fed. R. Crim. P. 11(d)–(e)).

In courts-martial, R.C.M. 910 requires the military judge to explain the rights waived, to determine personally that the plea is voluntary and not the result of force, threats, or promises outside a plea agreement, and to question the accused under oath to establish a factual basis. In this providence inquiry, the accused must admit every element and describe what he did (United States v. Care, 1969); the accused need not remember every circumstance but must be convinced of and able to describe the facts establishing guilt (R.C.M. 910(e), Discussion). If the accused sets up matter inconsistent with the plea, the judge must resolve the inconsistency or reject the plea, and inconsistent matter raised after findings requires a renewed inquiry into the plea’s providence (Article 45(a), UCMJ; R.C.M. 910(h)(2)). The pleas R.C.M. 910(a) allows are guilty, not guilty, guilty to a lesser included offense, and guilty with exceptions and substitutions; it makes no provision for nolo contendere or for a guilty plea accompanied by a claim of innocence. A guilty plea waives all motions and objections under Mil. R. Evid. 304 for that offense, except as provided for conditional pleas under R.C.M. 910(a)(2) (Mil. R. Evid. 304(f)(8)).

Arizona

Arizona permits pleas of guilty and no contest; a no contest plea may be accepted only after the court gives due consideration to the parties’ views and the public interest, and the court must determine that any plea is voluntary and intelligent and that there is a factual basis for it (Ariz. R. Crim. P. 17.1(b), (c), 17.3). The court may allow withdrawal of a plea when necessary to correct a manifest injustice (Ariz. R. Crim. P. 17.5). A defendant who pleaded guilty or no contest seeks post-conviction relief under Rule 33, which since January 1, 2020, has been separate from Rule 32 for defendants convicted at trial. Grounds include newly discovered material facts that probably would have changed the judgment or sentence (Rule 33.1(e)) and a claim of actual innocence shown by clear and convincing evidence that no reasonable factfinder would have found the defendant guilty beyond a reasonable doubt (Rule 33.1(h)).

Special Issues in Legal Practice

The research described above is the same whichever party relies on it. The science operates the same way regardless of who is asking the questions, and the research summarized here does not by itself favor either side. What differs between the parties is not the science but the vantage point — the particular places where these issues tend to surface, and the particular errors each party is best positioned to guard against. The two summaries below identify those recurring issues. They are offered for orientation, not as litigation strategy; in every case the controlling question is what the evidence in that case will, and will not, support, and the expert’s role is to inform the trier of fact rather than to advocate for either side.

Special Issues for Prosecutors

The research is consistent with the view that most guilty pleas are entered by people who committed the offense: in the laboratory studies reviewed here, guilty participants accepted offers far more often than innocent ones (Wilford & Wells, 2018; Henderson & Levett, 2018), and a confession’s effect on pleas largely reflects its evidentiary weight. Pleas also serve legitimate interests, including certainty of outcome and sparing complainants and other witnesses the burden of testifying. A careful, specific factual-basis inquiry is most informative when the defendant describes facts in his own words rather than agreeing to a summary. In courts-martial, the providence inquiry and Article 45 require that inconsistencies be resolved, which creates a record a reviewing court can evaluate.

The research also counsels caution. Large sentence differences, anchoring on a high threatened sentence, and offers that would end pretrial detention increased innocent participants’ willingness to plead (Kassin et al., 2025), and a false confession was the factor most strongly associated with false pleas among exonerations (Redlich et al., 2025). Where a case rests heavily on a disputed confession, an offer that depends on the defendant’s confinement status or that carries an unusually large discount may obtain a plea that does not resolve the underlying reliability question. Ruiz sets a constitutional minimum for disclosure before a plea; obligations under court rules and rules of professional conduct can be broader.

Special Issues for Defense

The research is relevant to how counsel assesses the reliability of a confession before advising a plea. Defense attorneys in hypothetical studies recommended pleas more often when a confession was admissible (Hellgren & Kassin, 2022), and a confession appears more often in false-plea exonerations than in trial exonerations (Cooper et al., 2019, as cited in Kassin et al., 2025; Redlich et al., 2025). Counsel should consult current authority on how a plea affects later challenges to a confession (see McMann; Tollett). Innocent participants were strongly influenced by an advocate’s recommendation (Henderson & Levett, 2018), which places weight on counsel’s assessment, and plea forms alone may not ensure understanding (Redlich & Bonventre, 2015).

The research also counsels caution. The laboratory studies show that most guilty participants accept favorable offers and that many innocent participants refuse them; they do not show that a client’s plea was false or that a plea offer was coercive. A client who maintains innocence but faces strong, corroborated evidence may reasonably prefer a plea, and courts-martial will not accept a plea from an accused who denies the elements. Post-conviction relief after a plea requires newly discovered evidence or clear and convincing proof of innocence in Arizona (Rule 33.1(e), (h)), and general research on false pleas does not meet that standard by itself.

This paper is for educational purposes only and is not legal advice. The author is a psychologist, not an attorney; summaries of law are general and must be verified against current authority or with a licensed attorney.

References

Dervan, L. E., & Edkins, V. A. (2013). The innocent defendant’s dilemma: An innovative empirical study of plea bargaining’s innocence problem. Journal of Criminal Law and Criminology, 103(1), 1–48.

DiFava, R. J., Bettens, T., Wilford, M. M., & Redlich, A. D. (2024). Confession evidence results in more true and false guilty pleas than eyewitness evidence. Journal of Experimental Criminology, 20(4), 1253–1267. https://doi.org/10.1007/s11292-023-09577-7

Fessinger, M. B., & Kovera, M. B. (2022). From whose perspective? Differences between actors and observers in determining the voluntariness of guilty pleas. Law and Human Behavior, 46(5), 353–371. https://doi.org/10.1037/lhb0000501

Hellgren, J., & Kassin, S. M. (2022). The defense lawyer’s plea recommendation: Disentangling the influences of perceived guilt and probability of conviction. Psychology, Public Policy, and Law, 28(4), 546–559. https://doi.org/10.1037/law0000365

Henderson, K. S., & Levett, L. M. (2018). Investigating predictors of true and false guilty pleas. Law and Human Behavior, 42(5), 427–441. https://doi.org/10.1037/lhb0000297

Kassin, S. M., Cleary, H. M. D., Gudjonsson, G. H., Leo, R. A., Meissner, C. A., Redlich, A. D., & Scherr, K. C. (2025). Police-induced confessions, 2.0: Risk factors and recommendations. Law and Human Behavior, 49(1), 7–53. https://doi.org/10.1037/lhb0000593

Redlich, A. D., & Bonventre, C. L. (2015). Content and comprehensibility of juvenile and adult tender-of-plea forms: Implications for knowing, intelligent, and voluntary guilty pleas. Law and Human Behavior, 39(2), 162–176. https://doi.org/10.1037/lhb0000118

Redlich, A. D., Summers, A., & Hoover, S. (2010). Self-reported false confessions and false guilty pleas among offenders with mental illness. Law and Human Behavior, 34(1), 79–90. https://doi.org/10.1007/s10979-009-9194-8

Redlich, A. D., Wilford, M. M., DiPano, M., & Berger, N. (2025). Commonalities in false guilty plea cases. Psychology, Crime & Law, 31(1), 64–82. https://doi.org/10.1080/1068316X.2023.2213381

Redlich, A. D., Yan, S., Norris, R. J., & Bushway, S. D. (2018). The influence of confessions on guilty pleas and plea discounts. Psychology, Public Policy, and Law, 24(2), 147–157. https://doi.org/10.1037/law0000144

Wilford, M. M., & Wells, G. L. (2018). Bluffed by the dealer: Distinguishing false pleas from false confessions. Psychology, Public Policy, and Law, 24(2), 158–170. https://doi.org/10.1037/law0000165

Legal Authorities

Bordenkircher v. Hayes, 434 U.S. 357 (1978).

Boykin v. Alabama, 395 U.S. 238 (1969).

Brady v. United States, 397 U.S. 742 (1970).

Lafler v. Cooper, 566 U.S. 156 (2012).

McMann v. Richardson, 397 U.S. 759 (1970).

Missouri v. Frye, 566 U.S. 134 (2012).

North Carolina v. Alford, 400 U.S. 25 (1970).

Tollett v. Henderson, 411 U.S. 258 (1973).

United States v. Ruiz, 536 U.S. 622 (2002).

United States v. Care, 40 C.M.R. 247 (C.M.A. 1969).

Uniform Code of Military Justice art. 45, 10 U.S.C. § 845.

Fed. R. Crim. P. 11.

Ariz. R. Crim. P. 17.1, 17.3, 17.5, 33.1.

Manual for Courts-Martial, United States, R.C.M. 910; Mil. R. Evid. 304 (2024 ed.).

Related topic papers: Risk Factors for False Confessions: An Overview · Corroboration and Assessing the Reliability of a Confession · Voluntariness and Suppression of Confessions · Confessions at Trial and After Conviction · Confessions in Courts-Martial: Article 31(b), Mil. R. Evid. 304 and 305, and Expert Assistance

Related topics from Arizona Forensic Services: Courts-Martial · Autism Online Offense