.png)
Should I retake the LSAT: it does not look bad, because schools work from your high score and 49.1 percent of 2024-25 test takers sat more than once. The real question is expected gain, which averages 2.39 points on a second attempt and 2.03 on a third.
No, and the structure of the process is why. Schools work from your high score, so an additional attempt that goes badly mostly sits beneath the score you already have.
It is also ordinary. In the 2024-25 testing year 49.1 percent of test takers tested more than once, per LSAC Technical Report TR 26-01 (https://www.lsac.org/sites/default/files/research/TR-26-01.pdf). Roughly half the field retakes, which makes a single attempt the minority experience rather than the norm.
The version of this that does read badly is not the number of attempts. It is several attempts with no movement, which tells a committee that the preparation is not changing. That is a fair inference and it is the thing to prevent.
Number of attemptsShare of 2024-25 test takersOnce only50.9 percentTwice29.7 percentThree times13.1 percentFour times4.7 percentFive or more1.6 percent
About two and a half points on average for a second attempt, and about two for a third. Mean second-attempt gain was 2.39 points in 2024-25, 2.50 in 2023-24 and 2.69 in 2022-23, and mean third-attempt gain was 2.03 points in 2024-25, all per LSAC Technical Report TR 26-01 (https://www.lsac.org/sites/default/files/research/TR-26-01.pdf).
Those numbers are the whole argument on this page, and they cut both ways. Two and a half points is enough to move a file across a school's median and is nowhere near enough to rescue a plan that needs eight.
AttemptMean point gainSecond attempt, 2024-252.39Second attempt, 2023-242.50Second attempt, 2022-232.69Third attempt, 2024-252.03
Set the question up as expected gain against expected cost, in that order, because most people do it in the reverse order and reach a decision by feeling.
Expected gain starts at the LSAC-published population mean of roughly 2.4 points for a second attempt, from the technical report cited above, and moves from there for reasons you can name. It moves up if your first sitting was disrupted, if you were systematically running out of time rather than getting questions wrong, or if you have not yet done the review method that produces gains. It moves down if you prepared properly, finished comfortably, and scored at your practice average.
Expected cost is a calendar cost more than a money cost. A later sitting means a later application, and at schools reading on a rolling basis that means competing for fewer seats and a more committed scholarship budget.
Your situationExpected gainUsual callDisruption on the day, illness or a center problemWell above the meanRetakeMistimed a section and left questions blankAbove the meanRetake, after fixing pacing specificallyScored at or above your practice average, finished comfortablyAt or below the meanUsually keep and applyScored well below practice average with no clear causeAround the meanRetake once, with a changed methodThird attempt, first two within two points of each otherBelow the meanThe method is the problem, not the attempts
Pages routinely claim that some share of retakers improve, some stay flat and some decline. LSAC does not publish that distribution in TR 26-01, and we could not source it anywhere authoritative (LSAC Technical Report TR 26-01 (https://www.lsac.org/sites/default/files/research/TR-26-01.pdf)).
Means are what exist. Treat any distribution claim as decoration, and be slightly more skeptical of the page that printed it.
Changing the method rather than adding hours. Across 28 studies and more than 16,000 participants, test prep produced an overall effect of g = .26, and neither contact hours at p = .489 nor program duration at p = .314 predicted outcomes, per Hao and colleagues (2025), Review of Educational Research (https://journals.sagepub.com/doi/10.3102/00346543251360775).
Practice tests alone came in at g = .19 and were not statistically significant at p = .233. Structured work on the reasoning did better: workbooks at g = .40 and explicitly teaching test-taking skills at g = .35, from that same Hao analysis. Sitting more timed tests is the most popular retake plan and close to the least effective one.
And spread the work out. Spaced practice produced 47.3 percent correct against 36.7 percent massed, with 259 of 271 comparisons favoring spacing, per Cepeda and colleagues (2006), Psychological Bulletin (https://www.yorku.ca/ncepeda/publications/CPVWR2006.pdf).
LSAC sets limits per testing period and over a lifetime. Our verified fact set does not include the current figures, so confirm them with LSAC directly rather than from any summary.
What the published data shows is behavior rather than limits: 50.9 percent of 2024-25 test takers sat once, 29.7 percent twice, 13.1 percent three times, 4.7 percent four times and 1.6 percent five or more, per LSAC Technical Report TR 26-01 (https://www.lsac.org/sites/default/files/research/TR-26-01.pdf).
For nearly everyone the limit is not the binding constraint. The calendar is.
Every LSAC figure on this page was taken from LSAC's own pages and last checked on July 26, 2026. Prices, dates and policies move without notice, so the source link sits next to each number rather than in a bibliography at the bottom.
Where LSAC has not published something, this page says so in those words rather than filling the hole with a plausible number. That is the difference between a gap and a guess, and only one of them is safe to plan around.
No. Schools work from the high score and roughly half the field retakes: 49.1 percent of 2024-25 test takers tested more than once, per LSAC. What reads badly is several attempts with no movement.
Mean second-attempt gain was 2.39 points in 2024-25 and mean third-attempt gain was 2.03 points, per LSAC Technical Report TR 26-01.
LSAC sets per-period and lifetime limits, which we did not verify and do not state here. In practice 50.9 percent of test takers sit once and only 1.6 percent sit five or more times.
LSAC does not publish it. Any figure circulating online is unsourced, so plan against the published means instead.
On the evidence, barely. Practice tests alone produced g = .19 at p = .233, which is not significant, while workbooks reached g = .40, per Hao and colleagues (2025). Changing the review method beats adding volume.
Written by Ali, Georgetown Law, founder of Lovare Institut.
August 5, 2026
August 5, 2026