Settlement Amounts by State — Cross-Source Percentile Benchmarks from 578,245 Public Payout Records (1990–2026)

State-level percentile benchmarks (p25, median, p75, p90) for settlement and claim payouts in all 50 US states + DC, computed from 578,245 public payout records across 9 government and municipal data sources, 1990–2026. The dataset combines National Practitioner Data Bank medical-malpractice payment reports (79.5% of records), municipal claim payouts published by New York City, Los Angeles, Chicago, Philadelphia, and Montgomery County (MD), California Prop 65 settlements and judgments, and the FiveThirtyEight police-misconduct settlement dataset. Four CSV tables are provided: percentiles by state, state-by-source composition, national by-source summary, and a national headline summary including all exclusion counts. Key figures: National median payout $72,500 (mean $214,059; p25 $15,000; p75 $245,000; p90 $545,000). Pennsylvania shows the highest state median ($195,000; 92.8% med-mal records); New York and California the lowest ($27,500), driven by large volumes of small municipal and consumer claims alongside their malpractice reports. Source composition is disclosed per state because it is the first-order driver of cross-state differences: 76,015 New York City municipal claims (median $10,000) sit next to New York's 60,690 medical-malpractice records (median $145,000). Cleaning rules: two unreliable sources are excluded entirely (DOJ press-release text-extracted amounts and Phoenix sentinel placeholder amounts; 28,627 rows), 3,217,941 rows without a valid 50-state+DC code are excluded via an explicit whitelist (federal payments without a state, territories, military codes, degraded mine-safety state codes, blanks), and all amounts must be positive. Percentiles use Python's statistics.quantiles with the deterministic 'inclusive' method. Full analysis: https://settlementinsight.com/research/settlement-amounts-by-state Caveat: Rows are individual payment/claim records, not cases or persons; state-to-state comparisons reflect source composition as much as underlying payouts. NPDB amounts are public-use-file range midpoints. These are descriptive statistics of public records — not predictions of individual case values.