The Engine

The treadmill runs on a rule change that’s now 20 years old, plus a test policy that was always going to end up here once enough people figured out how to use it. No bad intentions required.

LSAC lets you take the LSAT 5 times in any 5-year period, and 7 times over your lifetime.¹ It costs about $250 per sitting.² Those limits were meant to be fair, and on their own they’re fine. The problem is what happens to the results once you have them.

In June 2006, the ABA changed its reporting rule.³ Before that, law schools reported the average of a student’s LSAT scores. After it, they reported only the highest. The stated reason was to stop penalizing people who took the test more than once, which sounds reasonable. But it also meant retaking was now purely good for a school’s reported median, and so for its US News ranking. Once the rule changed, every school had the same incentive. Encourage retaking. Admit people on their highest score. Report the high scores. Every school runs the same math, they all report personal bests instead of typical scores, and the medians drift up on their own.

The first-time taker data tells the story.

From 2010 to 2018, about 68% of LSAT test-takers were sitting for the first time.⁴ That dropped to about 56% between the 2018-19 and 2022-23 cycles.⁴ By 2023-24, first-time takers were 48.6% of the total, and repeat takers (people who had taken it before) were 51.4%.⁴ More than half of everyone who sat for the LSAT in 2023-24 had done it before.

Think about what that flip means. In the early 2010s, most people in a testing room were there for the first time, so the score spread mostly reflected a broad, fresh sample of ability. By 2023-24, most people in the room had already done this at least once. They were back because their first score wasn’t good enough. That room’s scores are a large-scale hunt for everyone’s best day.

That matters because of what a best-day system does to the numbers. A single test score is your real ability plus or minus a few points of noise, since no test is perfectly reliable. Take it once and the noise averages out across a big group. Take it 4 or 5 times and keep only your highest result, and you’re reporting your ability plus your luckiest day.

Second-time takers gain about 2.6 points on average.⁵ That’s real. Part is genuine learning, part is getting used to the format, and part is good luck after a bad first day. Now multiply that across tens of thousands of people all reporting personal bests. School medians rise even though ability doesn’t.

James took the LSAT 4 times: 165, 170, 167, 170.⁶ His average across all 4 was about 168. His highest was 170. Schools see the 170, which is about 2 points above what he’d score on a typical day. Not much for 1 person. But run the same thing across a whole applicant pool, and a school’s reported 75th percentile stops being the score that 75% of the class fell below on a typical day. It becomes the score that 75% of them hit at their peak, which is a higher number.

The school then defends that median in the rankings by setting scholarship lines at or above it. The applicant who would have sat well above the 2022 median at a ranked-30s school now sits at or below the 2025 version. The median climbed while he was still prepping. And so he retakes, and the cycle runs again.

The 2006 rule change is the hinge. Before it, taking the test again was a wash for schools, because the average roughly tracked what the person actually scored across tries. After it, retaking became a no-downside bet for applicants and a median-inflating bet for admissions offices. Neither side made a bad individual choice. The structure produced a group result that no one had to vote for.

It also produced something the schools probably didn’t intend: a student body that looks better on paper than it performs in class. Say a school’s admitted class has a reported median of 166, but half those students’ typical score is really 163 or 164. That class will perform like a 163-164 class in 1L. When bar results come back weaker than the entering numbers predicted, the school has to explain the gap. Some do it honestly. A lot don’t. That gap gets its own chapter (Chapter 15). For now, hold the shape. The reported number overstates what to expect, and the students carry the cost.

There’s a useful comparison from a different test. The SAT was recentered (recalibrated) in April 1995 after years of score drift. One signal: the number of perfect 1600s jumped from 25 in all of 1994 to 137 on the recentered April 1995 test alone.⁷ Perfect scores don’t multiply by 5 in a year because a generation suddenly got smarter. The reset produced the jump, and the size of it shows how far the old calibration had drifted. The LSAT hasn’t been recentered, and it’s not in immediate danger of it. But the accommodations data and the repeater data are the same kind of signal. Scores at the top of the scale are rising faster than ability at the top of the scale.

I want to be fair about what this does and doesn’t mean. Retaking isn’t cheating. LSAC allows it, and schools ask for it. If you prep harder and score better on your 3rd try, that’s real. The 2.6-point average gain for second-time takers includes a genuine learning piece. And in Chapter 20 I’ll argue directly that building a genuine, repeatable score is the right move, and the only one that actually pays off.

A score you can hit consistently, day after day, under pressure, is an accurate credential. A best day out of 4 tries is a ceiling, and it’s hard to live under a ceiling you don’t know is one. A 170 that sits 2 points above your floor doesn’t hold up through 1L, the bar, and the career after, especially when the school’s reported median is built from everyone else’s ceilings too.

What the data does mean is that the medians schools report run structurally higher than what those same students would score on any given typical day. The schools and the rankings both know this. Applicants who understand it can build it into their strategy. The ones who don’t just see a number and aim at it, not realizing the number is already running ahead of them.

James has a 170. Schools that advertise scholarship eligibility starting at 170 have moved that line to 171, or 172. Or they still say 170, but the actual awards went to students with 172s. He knows his own score, but nobody tells him the floor moved.

That’s the engine. Retakes plus the highest-score rule, compounding year after year since 2006, with a repeater population that’s now the majority, producing school medians that reflect everyone’s best day instead of anyone’s typical one.

Watch: The Dirty Secret Behind Law School Scholarships

The Dirty Secret Behind Law School Scholarships

This chapter is from Anyone Can Beat the Law School Collapse by Steve Schwartz, LSAT Coach and Founder of LSAT Unplugged. I've been coaching the LSAT since 2005.

Published July 28, 2026.