Plea bargaining (United States, 1970 to 2024)

About 97 percent of federal and 94 percent of state convictions come from guilty pleas, the Supreme Court has called plea bargaining legitimate and then regulated it through defense counsel, and critics and defenders still disagree about whether sentences after trial amount to a penalty.

The record

How common pleas are

Notes: FY2024 Figure 5 is a share of cases, FY2021 and FY2022 are shares of convictions, so the figures are close but not the same measure. Sources: USSC Sourcebook 2024, Figure 5; BJS, Felony Sentences in State Courts, 2006; Frye. All opened. The Sentencing Commission page for the Sourcebook gave the 61,678 case count but not plea shares; those came from Figure 5, whose text we matched to categories by position.

What the Supreme Court has said

| Date | Case | Holding and key language | Vote | |---|---|---| | 1970 | Brady v. United States, 397 U.S. 742 | A guilty plea is not invalid merely because it was entered to avoid a possible death sentence under the Federal Kidnaping Act. A plea is "a grave and solemn act to be accepted only with care and discernment." The Court described a "mutuality of advantage" in plea bargaining: defendants reduce exposure, and the State saves resources. It declined to hold a plea compelled "whenever motivated by" a desire for a lesser penalty. Justice White wrote. | Vote not stated on the page we opened | | 1971 | Santobello v. New York, 404 U.S. 257 | Plea bargaining is "an essential component of the administration of justice." When a plea rests significantly on a prosecutor's promise, "such promise must be fulfilled." The prosecutor had promised no sentence recommendation; a successor urged the maximum one-year term. Burger wrote; Douglas concurred; Marshall, with two others, concurred in part and dissented in part. | Count not stated | | 1978 | Bordenkircher v. Hayes, 434 U.S. 357 | No due process violation when a prosecutor, in plea talks, carries out a threat to bring more serious charges the defendant is plainly subject to. The defendant faced 2 to 10 years on an $88.30 forged instrument; he refused a five-year offer and received a life sentence under the habitual criminal law. The majority saw "no such element of punishment or retaliation" so long as the accused is free to accept or reject. Blackmun dissented: "Prosecutorial vindictiveness in any context is still prosecutorial vindictiveness." | 5 to 4 (Stewart) | | 2012 | Missouri v. Frye | Defense counsel's failure to tell a client of a formal plea offer can be ineffective assistance. The offer was a misdemeanor plea with a 90-day recommendation; it lapsed, and the client later received three years. The Court: plea bargaining "is the criminal justice system." (The Court was quoting a law review article.) | 5 to 4 (Kennedy; Scalia dissenting) | | 2012 | Lafler v. Cooper | When bad advice leads a defendant to reject a plea and go to trial, he must show a reasonable probability the plea would have been offered and accepted and that the result would have been less severe. Offer: 51 to 85 months. Trial result: 185 to 360 months. Remedy: the State must reoffer the plea, and the trial court has discretion. Scalia dissented, calling the remedy "unheard-of in American jurisprudence" and saying it creates a new field of "plea-bargaining law." | 5 to 4 (Kennedy) |

Sources: Cornell LII texts of Brady, Santobello, Bordenkircher, Frye and Lafler; all opened.

Sentence gaps between plea and trial

• In the BJS 2006 state data, mean total incarceration was 29 months for guilty pleas and 78 months for trial convictions; mean prison sentences were 47 months (plea) and 100 months (trial). Means exclude death and life sentences. The table does not hold offense type or criminal history constant, so it cannot isolate a "penalty." (BJS, 2006; opened)
• The National Association of Criminal Defense Lawyers (NACDL) 2018 report, as quoted in a 2023 coalition letter, found that for most offense categories "the average trial sentence in the federal system is three times higher than a plea sentence," and up to eight times for some crimes. We did not open the report; we read the letter only. The letter was signed by NACDL, the ACLU, Americans for Prosperity, FAMM, Right on Crime and The Sentencing Project, among others. (Coalition letter to USSC, Nov. 17, 2023; opened)
• The letter asks the Commission to let courts give a third acceptance-of-responsibility point after trial in the interests of justice and without a government motion (USSG 3E1.1(b)), to clarify that the obstruction adjustment (3C1.1) does not apply solely because a defendant testified, and to bar use of acquitted conduct. We did not open the Guidelines text.

Wrongful convictions and pleas

• The National Registry of Exonerations lists 3,867 exonerations since 1989. The Registry's guilty-pleas page, "Innocents Who Plead Guilty" (Nov. 24, 2015), reports that 15 percent of known exonerees pleaded guilty (261 of 1,702 at that time). A 2023 NACDL letter gives 18 percent; we could not match it to a Registry source, so we use the Registry's figure. (National Registry of Exonerations; opened)
• Judge Jed Rakoff of the Southern District of New York wrote in 2014 that the system is "almost exclusively a system of plea bargaining" and cited criminologists who estimate that 2 to 8 percent of convicted felons who plead guilty are innocent (as reported in a Marshall Project republication that also carries a judicial reply). (Marshall Project, Dec. 26, 2014; opened) The estimate is contested and we did not trace its source.

What this case shows

1. Most people never reach the public part of the system. With 97.2 percent of federal cases ending in a plea, a jury sees about 3 in 100. The public sees sentencing and charges; the offer, the deadline and the alternative charge sit in private talks. (USSC)
2. The rules were set by courts, not legislatures. Brady, Santobello and Bordenkircher are Supreme Court decisions; no statute in our sources defines plea bargaining. In 2012 two 5 to 4 decisions said defense counsel's advice during plea talks is covered by the Sixth Amendment. (Cornell LII)
3. A visible gap, an unsettled cause. State and federal figures show much higher average sentences after trial. The causes are disputed: offense seriousness, record and cooperation also differ. The coalition's proposed fixes (third acceptance point, limits on obstruction and acquitted-conduct enhancements) are specific and checkable. (BJS; NACDL letter)
4. Those with less counsel have less leverage. Frye and Lafler involved lawyers who failed in advising clients on offers; both turned on a lawyer's conduct, not on a defendant's choice. Those without effective counsel were therefore the people the Court had to protect. We found no data on how often this occurs. (Cornell LII)
5. Both sides accept that few cases can be tried. A Tulane philosophy professor writing in a 2015 reply to Rakoff estimates trying every case would take six or seven times as many judges; critics and defenders disagree about whether that is a reason to keep bargaining or to shorten trials. (NYRB exchange, opened)

The strongest objections, and our answers

1. Objection: Pleas are voluntary bargains that benefit both sides. The Court in Brady described "mutuality of advantage." Robert Scott and William Stuntz (Yale Law Journal, 1992) wrote that most scholars condemn plea bargaining yet courts and participants accept it, and that contract theory is a natural way to regulate it. Mitigation: We report this as the Court's and a leading scholarly defense. We did not read the full article, only an abstract.
2. Objection: There is no "trial penalty"; there is a discount for accepting responsibility and saving resources. Brady supports the idea that defendants who plead receive a reduction in exchange for saved cost. Bordenkircher held that carrying out a threatened charge is not punishment for exercising a right when the defendant is free to refuse. Mitigation: We state both holdings. The dispute is whether the gap in outcomes is a discount or a penalty, and the evidence we opened cannot separate them.
3. Objection: The averages compare different cases. Correct. The BJS table does not control for offense or record, and the NACDL ratio comes from a defense group's analysis of Sentencing Commission data. Mitigation: We say so and show the figures as raw means and as claims.
4. Objection: Abolition would clog courts and leave the innocent worse off. The reply by Bruce Brower cites caseloads; Stephen Schulhofer, an abolitionist scholar, disputes the claim that abolition would harm the innocent and says it would improve both justice and efficiency (abstract only). Mitigation: We show both sides and make no claim about which is right.
5. Objection: A single high-profile case (Bordenkircher) or a judge's view (Rakoff) is not the system. Mitigation: Correct. Rakoff's 2 to 8 percent is an estimate he attributes to criminologists. We label it unverified.

Open questions: what would change our minds

• How much of the plea-trial sentence gap remains after matching for offense, criminal history and cooperation? A study using Sentencing Commission microdata would decide this; we did not open one.
• The NACDL 2018 report itself and its method. The page was unreachable; we read only a later letter.
• The Registry's share of exonerees who pleaded guilty (15 percent) is from 2015; a current figure was not checked.
• Whether the 2 to 8 percent innocent-plea estimate has a traceable empirical source.
• The Sentencing Guidelines text for 3E1.1 and 3C1.1 was not opened.
• Vote counts in Brady and Santobello: not shown on the pages we opened.
• State practice after Frye and Lafler (how many offers are put on the record) not examined.
• Federal Rule of Criminal Procedure 11 and the Sentencing Reform Act's treatment of plea agreements: not opened.
• Whether defenders of prosecutorial practice (for example the National District Attorneys Association) have a published response; we searched but did not open one.

Sources

Evidence status and correction history

Key claims

Correction history

Found an error or a better source? We keep every correction on the record with its date and reason. Nothing here is quietly rewritten.