Qualified immunity (United States, 1967 to 2025)

Since 1967 the Supreme Court has shielded officials from money damages unless they broke "clearly established" law; defenders cite fair notice and fear of chilling officials, critics cite a gap in accountability, and a few states have changed the rule.

The record

What the doctrine is

• Qualified immunity is a rule made by courts, not written out in the text of Section 1983 (42 U.S.C. 1983), the federal law that lets people sue state and local officials for violating federal rights. (Described in Nielson and Walker, 2018; the dispute over whether the doctrine has a legal basis is Section 3 below.)
• Today's test comes from Harlow v. Fitzgerald (decided June 24, 1982). Officials performing discretionary functions are shielded from damages unless their conduct violates "clearly established statutory or constitutional rights." The Court named the social costs it was weighing: "the diversion of official energy from pressing public issues" and "the expenses of litigation." It dropped the earlier test of an official's state of mind, because that test "frequently has proved incompatible with" ending insubstantial claims before trial. (Justice, Harlow v. Fitzgerald, 457 U.S. 800; opened)
• It is both a defense to paying damages and, in the Court's words in Pearson, "an immunity from suit rather than a mere defense to liability." (Justice, Pearson v. Callahan, 555 U.S. 223; opened)

Timeline of the main court decisions

• In Pierson the Court wrote that officers are not required to predict future constitutional rulings. (Justice, Pierson v. Ray; opened)
• Hope and Taylor show that the Court has, in some cases, denied or vacated immunity without a case on identical facts. Taken together, the opinions show a rule that can protect officials in some cases and not in others, depending on how the lower court reads "clearly established." (Cornell LII, Hope; Cornell LII, Taylor; opened)

Empirical studies

• Joanna Schwartz, "How Qualified Immunity Fails," 127 Yale Law Journal (2017). She read dockets of 1,183 Section 1983 cases against state and local law enforcement filed in 2011 and 2012 in five federal districts. Qualified immunity could not be raised in 17.3% of them. It produced dismissals in 38 cases, which is 3.9% of the 979 cases where it could be raised, or 3.2% of all 1,183. She concludes it "rarely served its intended role as a shield from discovery and trial" in this sample and cites earlier work that law enforcement defendants are almost always indemnified (paid for) by their employers. She notes limits: five districts, two years, no state court cases, and no way to measure deterred filings or settlements. (Yale Law Journal; opened, last 14,000 characters not read)
• Aaron Nielson and Christopher Walker, "A Qualified Defense of Qualified Immunity," 93 Notre Dame Law Review 1853 (2018), respond. They say the sample is limited, that settlements and deterred suits are not captured, and that qualified immunity "mattered" in about 30% of cases in circuit court data they examine (circuit courts denied it on clearly established grounds in 28% of claims and did not reach the constitutional question in 27%). They also say policy objections belong to Congress. (Notre Dame Law Review PDF; opened, as summarized by our fetch tool)
• Reuters, 2020: In 529 published federal appeals opinions from 2005 to 2019 in excessive force cases where officers raised the defense, courts granted immunity in 44% of cases in the first three years and 57% in the last three. Reuters also counted 121 Supreme Court petitions on the issue from 2005 to 2018 and found the Court was 3.5 times as likely to grant review when the officer filed. A separate Reuters analysis of 252 appeals opinions from 2015 to 2019 found immunity granted in 57%, and about three dozen cases where a court found a violation yet granted immunity. (SCOTUSblog, Ask the author; Reason summary; both opened. We did not open the Reuters pages themselves.)

Changes by legislatures

• Colorado: SB 20-217 was signed June 19, 2020. It creates a state-law civil action against peace officers for violating the state bill of rights and says "Qualified immunity is not a defense to the civil action." Local governments must indemnify the officer, except that if the employer finds the officer did not act on a good faith and reasonable belief the act was lawful, the officer pays 5 percent of the judgment or $25,000, whichever is less. (Colorado General Assembly bill page; opened, summary only; the bill text was not opened)
• New Mexico: House Bill 4, the New Mexico Civil Rights Act, was signed in April 2021 (Reason's article is dated April 7, 2021 and does not state the date). It bars qualified immunity as a defense in state court civil rights claims and puts costs on the government, not the individual. Every Republican in the state House and five Democrats voted against it. (Reason; opened)
• Congress: the Ending Qualified Immunity Act (119th Congress, Senate version introduced by Mr. Markey) would add a subsection to 42 U.S.C. 1983 that bars defenses based on a defendant's good faith, belief the conduct was lawful, or that rights were not "clearly established." We did not confirm any floor vote or enactment. (Bill text PDF hosted by Rep. Pressley; opened)

Disputed or differing figures

| Item | Figure A | Figure B | Notes | |---|---|---| | Share of cases where immunity is granted | 3.2% of all cases (38 dismissals, 3.9% of cases where it could be raised), district courts, 2011 to 2012 (Schwartz) | 44% to 57% of excessive force appeals, 2005 to 2019 (Reuters) | Different courts, stages, years and case types, so not a contradiction. Appeals cases are the ones that reach appeal. | | How much it "mattered" | "rarely served its intended role" (Schwartz) | "mattered" in about 30% of circuit cases (Nielson and Walker) | Different data sets and different definitions of "mattered" | | Reuters appeals sample | 529 opinions, 2005 to 2019 | 252 opinions, 2015 to 2019 | Both reported by Reuters; the second is a subset period |

What this case shows

1. A rule written by judges, applied one case at a time, is hard for the public to see or vote on. The doctrine is not in the statute. Voters and legislators can change it (Colorado and New Mexico did), but most people learn of it only after a suit ends. (Nielson and Walker; Colorado bill page)
2. The accountability gap is documented in the case record. Reuters found about three dozen appeals where courts found a violation and still granted immunity. Courts also increasingly skip the first question after Pearson, which leaves fewer written rulings on what the law is. (Reason summary of Reuters; Pearson)
3. The cost of the rule falls on injured people, and the cost of changing it falls on local budgets. Schwartz reports officers are almost always indemnified, so employers and taxpayers pay. Colorado and New Mexico legislated who pays: the government, with a limited officer share in Colorado. (Yale Law Journal; Colorado bill page; Reason)
4. The justification for the rule has been questioned with data, and the data are themselves contested. One side says the shield rarely ends cases early; the other says the sample is small and cannot see deterred suits. The public rarely sees the data. (Yale Law Journal; Notre Dame Law Review)
5. The Court itself has sometimes denied immunity on obvious facts. Hope and Taylor show "clearly established" need not mean a prior case with the same facts, which means outcomes depend on how lower courts apply it. (Cornell LII)

The strongest objections, and our answers

1. Objection: Officials must act fast under unclear law, and personal liability would chill legitimate work. The Harlow Court cited the "diversion of official energy" and Nielson and Walker cite the risk of chilling officers. Mitigation: We state this as the main reason the doctrine exists. We note that Colorado kept most of the cost on employers, not officers, and that no study we opened measures a change in police conduct after the reforms.
2. Objection: Schwartz's study is too narrow to count. Five districts, two years, no state courts, and it cannot see deterred suits or settlements. Mitigation: We report those limits from her own article and show the competing 30% figure. We do not say who is right.
3. Objection: The doctrine is settled law and belongs to Congress. Nielson and Walker rely on precedent and say policy critiques belong to Congress. Mitigation: We include that. The doctrine's critics also argue it has weak legal grounding (Part I of Nielson and Walker responds to scholar William Baude on this point). We did not open Baude's article, so we do not summarize his case.
4. Objection: State reforms prove nothing yet. Mitigation: Correct. We found no outcome data from Colorado or New Mexico. We list the laws as changes in who can sue and who pays, not as evidence of better or worse policing.
5. Objection: Reuters and Reason are journalism and advocacy, not research. Mitigation: We use them for counts that can be rechecked and mark them secondary. We did not open the Reuters pages themselves.

Open questions: what would change our minds

• Outcome data from Colorado and New Mexico after the laws (filings, settlements, costs to local governments, police behavior). None opened. Unverified.
• Whether any federal bill to end qualified immunity passed either house. Not confirmed.
• The full Colorado and New Mexico bill texts (we read the Colorado General Assembly summary and a news article).
• William Baude, "Is Qualified Immunity Unlawful?" (California Law Review, 2018) was not opened. Our description of it comes from the Notre Dame article.
• Other empirical work (for example studies of state court cases or of deterrence) was not opened. A broader data set could change the Schwartz versus Nielson and Walker balance.
• The text of Hope v. Pelzer and Taylor v. Riojas was read through a short summary; the full opinions, including Justice Thomas's dissent in Taylor, were not read.
• Reuters' own methodology page was not opened.

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.