Picture a federal courthouse in 2026. Rows of courtrooms, most of them dark most of the time. The dockets are full. The trials are not happening. Roughly 97% of civil cases never see a jury — they settle, get dismissed, or resolve on motion. About 1% end in a trial verdict. The jury-trial disposition rate in federal court is approximately 0.7%.
It is the most quoted number in American litigation, and most people quote it without asking where it came from. The figure traces to a 2004 U.S. Department of Justice study of cases in the nation's 75 largest counties. Randall Kiser of the litigation-analytics firm DecisionSet later offered a range of 80 to 92% in the New York Times. Take either number. The conclusion does not change: trial is the rare exception, not the rule. And 97% is arguably conservative — it counts only filed cases and misses entirely the disputes settled before a complaint is drafted, a shadow population experienced litigators know is enormous.
The interesting question is not whether the number is exact. It is why the number is what it is.
The System Is Engineered to Produce Settlements
The American trial did not fade by accident. The 1938 Federal Rules of Civil Procedure created broad pretrial discovery, and that single reform did more to end the trial than any other development: by the time discovery closes, both sides usually know enough about the other's evidence to price the case within a narrow band. Summary judgment, mandatory mediation programs, court-annexed arbitration — each pushes toward the same exit.
Marc Galanter documented the collapse in a landmark 2004 study in the Journal of Empirical Legal Studies: federal civil trial rates fell from roughly 20% of dispositions in the 1930s to under 2% by the early 2000s. The trial did not become fairer or rarer by noble design. It became expensive, slow, and unnecessary for fact-finding — and the system adapted around it.
Criminal justice reached the same destination by a different road. As Justice Anthony Kennedy wrote for the Supreme Court in 2012, criminal justice is "for the most part a system of pleas, not a system of trials" — 97% of federal criminal convictions come by guilty plea. The civil system got there through discovery; the criminal system through prosecutorial leverage. Both converged on the same arithmetic.
Certainty Has a Price, and Settlement Is How the Market Clears It
Strip away the procedure and every settlement is a pricing decision. Each case has an expected trial value — a probability-weighted outcome. Both sides discount it: the plaintiff for the risk of losing outright and the years of waiting, the defendant for the risk of a runaway verdict and the certainty of legal fees. When the discounted valuations overlap, the case settles. When they don't, somebody has mispriced risk.
The numbers make it concrete. Filing to verdict in a personal injury case: two to four years. Expert costs in a medical malpractice case: $15,000 to $30,000 before trial, and far higher in complex commercial disputes. A defendant facing plausible seven-figure exposure will rationally pay six figures to extinguish a 20% chance of catastrophe. That is not weakness. It is arithmetic.
Then there is the information dynamic. Uncertainty shrinks as litigation advances — the deposition that goes sideways, the expert who gets excluded, the summary-judgment order that narrows the claims. Each event reprices the case, and settlements cluster around these inflection points because that is where both sides' valuations converge. The best litigators do not prepare for trial in the abstract. They manufacture the events that force repricing.
When Cases Actually Settle
The timing of settlement is not random, and it follows a pattern every experienced litigator recognizes. A meaningful share of cases resolve before discovery even begins — the demand letter, the early mediation, the defendant who reads the complaint and does the math. These are the cheapest settlements for both sides and, for plaintiffs' attorneys working on contingency, the most profitable per hour invested.
The second wave comes after key depositions. Once the plaintiff has testified and the defendant's witnesses have been examined, the information asymmetry collapses. Both sides have seen the other's case at its most vulnerable. Settlement authority that was unavailable six months earlier suddenly materializes.
The third wave is the courthouse steps — the days and weeks before trial, when sunk costs are highest and the uncertainty is most acute. This is when the largest settlements happen, because the alternative is no longer theoretical. A trial date concentrates the mind like nothing else in litigation.
What almost never happens is the mid-discovery settlement for full value. Cases in the thick of discovery, with costs mounting and no inflection point in sight, are where valuation gaps are widest. The lesson for anyone pricing a dispute: value is a function of time as well as merit, and the same case is worth different amounts at different stages.
The Warning Hidden in the Statistic
Here is what the 97% figure actually tells a sophisticated observer: if you value a case by asking "what would a jury award," you are modeling an event with a 1% probability. The real question — what does this case settle for — is a function of leverage, timing, cost, and the opponent's risk tolerance. The merits matter, but they are one input among several.
This is why settlement databases are more valuable than verdict reporters for pricing, why mediators with deep market knowledge command premium fees, and why sophisticated defendants maintain settlement guidelines calibrated to case type and venue instead of relying on ad hoc judgment.
"The trial is the theater. The settlement is the market — and markets price on data, not drama."