If you are comparing US law schools, the single most useful public dataset is the ABA 509 disclosure. Every ABA-accredited school publishes one each year, and it contains the numbers that actually matter for admissions: acceptance rate, the median LSAT and GPA of the entering class, and the 25th/75th percentiles behind those medians.

Why the 509 is worth reading

Most ranking systems mix in reputation surveys or outcome data that are hard to reproduce. The 509 is different: the figures are reported by the schools themselves to the ABA, so you can verify them at the source. A few things I always look at:

  • Median vs. 25th/75th percentiles. The median LSAT tells you the midpoint, not the floor. If your score sits below the 25th percentile, the median alone gives a false sense of fit.
  • Acceptance rate in context. A 40% acceptance rate at a highly selective school and a 40% rate at a regional school mean very different things once you account for who applies.
  • Cost and aid together. Sticker tuition is misleading without the grant data. The schools that publish both give a much clearer picture of the real three-year cost.
  • Bar passage, not just placement. For a professional degree, ultimate bar passage rate is a better signal than a broad employment headline.

A note on data quality

One trap I hit repeatedly: some fields are reported per credit hour and others as a full annual total. Adding them together produces nonsense. Any tool that normalizes this data should keep the two units separate and label each row with what was actually reported.

I built OpenLawPrep as a free, no-signup way to browse these disclosures across all 196 ABA law schools — acceptance rates, LSAT/GPA medians, tuition, grants and bar passage, plus a few simple calculators. It is just one interface over public data; the underlying 509 forms are the real source of truth.

If you are building something with admissions data yourself, my advice is simple: keep every figure traceable to its disclosure year, and never silently fill a missing value with zero.