Who Voted With Whom
Five Courts, 278 pairs of justices, and every contested case since 1946. The blocs are real, and they are older than you would guess.
Brennan and Marshall are the closest pair in the modern era — the same side of the judgment in 93 percent of contested cases. Rehnquist and Douglas are the furthest apart, at 13. On the Court sitting now the tightest pair is Roberts and Kavanaugh. And the whole exercise only works if you throw out the unanimous decisions first: 40 percent of the Court’s cases have no dissent at all, and counting them makes everyone look like friends.
What is being measured
Two justices agree, for this purpose, when they land on the same side of the judgment — both in the majority, or both in dissent. Every number below counts only divided cases: cases in which at least one justice dissented. That exclusion is the whole reason the matrices say anything. Unanimous decisions are 40 percent of the modern docket, and a measure that counts them puts 85 percent of all pairs above 60 — including justices who disagree about nearly everything contested.
Agreement on the outcome is not agreement on the law. A justice can join a judgment and write separately to say the majority reached it for the wrong reasons; that is scored here as agreement. Two justices can dissent from the same decision for opposite reasons and be scored the same way. Within each matrix the justices are ordered most to least liberal for that Court, which is what makes the blocs fall into the corners — but no number depends on that ordering.
The Closest
The closest pair of the modern era
Rehnquist Court · 361 divided cases · 95.8% across all 577 they shared
After the Court reinstated capital punishment in Gregg v. Georgia (1976), Brennan and Marshall dissented from death sentences as a standing practice, each with his own recurring formula. Brennan’s ran: “Adhering to my view that the death penalty is in all circumstances cruel and unusual punishment prohibited by the Eighth and Fourteenth Amendments, Gregg v. Georgia, 428 U.S. 153, 227 (1976), I would grant certiorari and vacate the death sentence in this case.” Carol Steiker, who clerked for Marshall, has recalled typing that dissent over and over. It is the most literal illustration of two justices voting as one, and it is only part of why they finish first here.
The closest pair on the current Court
Roberts Court · 251 divided cases · 94.7% across all 433 they shared
This one has an independent check. SCOTUSblog’s own Stat Pack, built from a different source and a slightly different definition, put Roberts and Kavanaugh at 91 percent — against the 90.8 computed here — and noted the pair had aligned more often than any other in previous terms. Worth a caveat: single terms move around, and in the 2024 term alone SCOTUSblog’s highest pair was Thomas and Alito at 97 percent.
The Chief and his strategist
Warren Court · 1,168 divided cases · 92.3% across all 1,825 they shared
Brennan is remembered less as Warren’s follower than as the tactician of the Warren Court’s jurisprudence — his Baltimore Sun obituary in 1997 called him, more than any justice, “the strategist behind Supreme Court jurisprudence.” The 88 percent here is across 1,168 divided cases, which is a very long time to agree with anyone.
Two originalists, one difference
Rehnquist Court · 691 divided cases · 92.6% across all 1,254 they shared
The gap between them is not about the Constitution’s meaning but about what to do with precedent that got it wrong. Asked what weight stare decisis carried for him, Thomas answered: “Not enough to keep me from going to the Constitution.” Scalia was the more willing of the two to leave a mistaken precedent standing — which is where their remaining 13 percent of divided cases tends to live.
The Furthest Apart
The widest gulf in the database
Burger Court · 365 divided cases · 43% across all 560 they shared
They overlapped for less than four years — Rehnquist took his seat in January 1972, Douglas retired that November of 1975 — and in the 365 divided cases they shared they landed on the same side 46 times. Martin-Quinn scores, the standard scaling of judicial ideology, place these two at the empirical extremes of the whole modern Court. We found no account of any personal quarrel; the distance appears to be entirely a matter of law.
The long version of the same gulf
Burger Court · 1,443 divided cases · 50.8% across all 2,289 they shared
Rehnquist and Douglas is the widest pair, but it rests on a short overlap. Marshall and Rehnquist sat together for nineteen years in all — about fourteen of them under Burger, where they shared 1,443 divided cases and agreed in 21.9 percent, and the remainder under Rehnquist himself, where they agreed in 19.4. That is the most durable disagreement in the modern record.
Five Courts
Each cell is the share of divided cases in which that pair landed on the same side. Justices run most to least liberal, top to bottom and left to right, so the blocs sit in the top-left and bottom-right corners. A dash means the two shared fewer than 30 divided cases. Hover any cell for the counts behind it.
The Vinson Court
OT1946–OT1952 · 789 cases, 535 of them divided · 11 justices with at least 150 cases
The Warren Court
OT1953–OT1968 · 2,175 cases, 1,374 of them divided · 16 justices with at least 150 cases
The Burger Court
OT1969–OT1985 · 2,772 cases, 1,705 of them divided · 13 justices with at least 150 cases
The Rehnquist Court
OT1986–OT2004 · 2,029 cases, 1,142 of them divided · 14 justices with at least 150 cases
The Roberts Court
OT2005–OT2024 · 1,460 cases, 789 of them divided · 15 justices with at least 150 cases
Method, and its limits
Built from the Supreme Court Database (2025 Release 01, justice-centered), covering the 1946 through 2024 terms. One vote per justice per case. A pair enters a matrix only if both justices have at least 150 cases under that Chief; a cell is left blank below 30 shared divided cases. Cases where the database records a justice on both sides, or gives no determinate side at all — no recorded vote, or an equally divided Court — are excluded rather than guessed at.
Grouped by Chief Justice, not by natural court. The database marks 36 distinct natural courts in this period, each a stretch with unchanged membership, and that is the statistically cleaner unit: every pair inside one shares the same denominator. It is also unreadable — the median natural court is 160 cases, and they have no names anyone knows. Grouping by Chief buys legibility at the price of uneven denominators inside a matrix, since membership turns over underneath a long Chief. Brennan and Marshall’s Burger-era cell rests on 1,632 divided cases; others in the same grid rest on a few hundred. The counts are on every cell.
One outside check. SCOTUSblog publishes agreement tables each term in its Stat Pack, built from its own tally rather than from this database. For the 2024 term it put Roberts and Kavanaugh at 91 percent; the figure computed here is 90.8. The definitions are close but not identical — SCOTUSblog’s broadest tier counts agreement in full, in part, or in the judgment only — so treat the match as reassurance rather than proof.
Sources for the pairs above: Brennan’s recurring capital dissent as it appears in McCray v. Florida, 454 U.S. 1041 (1981), citing Gregg v. Georgia, 428 U.S. 153, 227 (1976) — Marshall used his own wording for the same practice; Carol Steiker on typing it, via the Brennan Center for Justice. Brennan as strategist, from his Baltimore Sun obituary, 1997. Thomas on stare decisis, “not enough to keep me from going to the Constitution.” Rehnquist took his seat 7 January 1972; Douglas retired 12 November 1975.
Questions
Which Supreme Court justices vote together most often?
William Brennan and Thurgood Marshall, by a clear margin — they landed on the same side of the judgment in 93.4% of the divided cases they shared on the Rehnquist Court, and 86.8% across 1,632 divided cases under Chief Justice Burger. On the current Court the closest pair is John Roberts and Brett Kavanaugh at 90.8%, a figure SCOTUSblog independently puts at 91%.
Which two justices disagreed the most?
William Rehnquist and William O. Douglas, who agreed in only 12.6% of the 365 divided cases they shared before Douglas retired in 1975. Over a much longer overlap — nineteen years, spanning two Chief Justices — Thurgood Marshall and Rehnquist agreed roughly a fifth of the time, which is the most durable disagreement in the modern record.
Do justices agree more often than people think?
Yes. Across the 9,225 cases here, 40% were decided without a single dissent. That is why this page reports divided cases separately: counting unanimous decisions makes every pair on the Court look like allies, including justices who disagree about nearly everything that is actually contested.
Does agreeing on the outcome mean agreeing on the law?
No, and this is the main limit of the measure. A justice can join the judgment and write separately to say the majority reached it for entirely the wrong reasons — that counts as agreement here. Two justices can also dissent from the same decision for opposite reasons and be scored as agreeing. The number tracks which side of the judgment each landed on, nothing more.
Companion pages: how far each justice moved across a career, and every justice’s voting record by issue area on their own profile.