Caselaw Access Project: 6.7 Million U.S. Court Cases Spanning 365 Years

6.7 million U.S. federal and state court decisions spanning 365 years, compiled from the Caselaw Access Project and Court Listener. This collection aggregates nearly 40 million pages of judicial opinions and decisions from 479 distinct courts. The dataset was curated by common-pile and includes both digitized historical records and modern scraped legal data.

Use Cases

• Training large language models for legal text generation using full-text judicial opinions
• Longitudinal analysis of U.S. legal trends and judicial language over three centuries
• Extraction of legal citations and precedent networks across state and federal jurisdictions

Strengths

• 6.7 million total case records
• 365-year temporal coverage of the U.S. legal system
• Aggregates data from 479 distinct courts
• Combines digitized historical pages with modern web-scraped records

Limitations

• Potential for OCR inaccuracies in historical records from the 40 million digitized pages
• Inconsistent formatting likely due to merging two distinct primary sources (CAP and Court Listener)
• Geographic bias restricted exclusively to United States jurisdictions

Provenance

Source

Caselaw Access Project and Court Listener

Collection Method

Digitization of physical court records and web scraping of 479 court websites

Time Range

1660-2025

Freshness

Updated as of June 2025; includes historical data dating back approximately 365 years.

Geography

United States