.png)
The honest answer to what LSAT score do I need for law school is a table, not a number: roughly 173 and up for the top of the T14, 168 to 172 for the rest of it, 163 to 167 for strong regionals, and 155 to 162 for most ABA schools. Here is how to read yours.
Set the target from the enrolled class, not from the brochure. Every ABA school publishes its 25th, 50th, and 75th percentiles annually, and those three numbers are the entire basis for a rational target.
TierLSAT bandGPA bandWhat it takes you toTop of the T14173 to 1753.92 to 3.99Yale, Harvard, Chicago, Stanford, Columbia, Penn, Virginia, NorthwesternRest of the T14170 to 1723.88 to 3.95NYU, Duke, Michigan, Berkeley, Cornell, UCLA rangeStrong regional and T30163 to 1693.6 to 3.9Georgetown at 171, George Mason 169, GW 168 as worked examplesBroad ABA middle155 to 1623.3 to 3.7American at 162 as a worked example; most ABA schools
T14 bands from each school's 2025 ABA 509 report; named school figures verified from the 2025 509s on July 26, 2026. Within the T14, the lowest median LSAT was 170 and the highest 174.
The bands overlap deliberately, because schools overlap. A 169 is a median at George Mason, a reach at NYU, and a strong scholarship number at American, all in the same cycle.
A scaled score means nothing until you know what share of test takers it beats, and the scale is not linear in difficulty. LSAC publishes the conversion.
LSATPercentilePoints of percentile gained from 5 more points15038.06%+17.9 to reach 15515555.91%+17.0 to reach 16016072.92%+13.6 to reach 16516586.50%+8.6 to reach 17017095.07%+3.8 to reach 17517598.91%+1.0 to reach 180
Source: LSAC percentile table covering the 2022-2023 through 2024-2025 testing years, https://www.lsac.org/data-research/data/lsat-percentiles. The right column is our arithmetic from the same table.
Read the right column carefully, because it governs whether studying more is worth your time. Five points is worth 17.9 percentile points at 150 and 1.0 at 175.
The practical rule that follows: below 165, additional points are the highest-return investment available to you in the entire application. Above 172, the marginal point buys much less, and your time is usually better spent on the file and the school list.
It is slightly below the national median and it is workable almost everywhere with the right LSAT. Fall 2025 entering classes across all ABA schools clustered near a 3.65 median GPA (est: per LSD.Law aggregation of ABA 509 reports, https://lsd.law/law-school-lsat-gpa-gre), so a 3.5 sits just under the middle of the national distribution.
What matters more is where it sits at your specific targets. A 3.5 is above American's 3.32 twenty-fifth percentile and below its 3.63 median, comfortably inside its admitted range; at Georgetown, whose 25th percentile is 3.75, a 3.5 is below the bottom quarter of the class and needs the LSAT to carry the file.
SchoolGPA 25thGPA medianWhere a 3.5 sitsGeorgetown3.753.93Below the 25th: splitter territoryNYU3.813.92Below the 25th: splitter territoryGeorge Mason3.553.93Just below the 25thGW3.553.86Just below the 25thAmerican3.323.63Between the 25th and the median
All figures from each school's 2025 ABA 509 report, verified July 26, 2026.
Note something the table makes obvious: Georgetown and GW have nearly identical GPA medians of 3.93 and 3.86, but their 25th percentiles differ by 0.20. The floor, not the median, is what tells you whether a school takes applicants like you.
A splitter is an applicant whose LSAT sits at or above a school's median while their GPA sits at or below its 25th percentile. A reverse splitter is the mirror image: strong GPA, weak LSAT.
Splitters exist because schools report both medians to the ABA and both are scarce. An applicant who delivers a high LSAT gives the school the number that is hardest to buy, so schools will accept a GPA hit to get it.
The disclosed spreads are the evidence. When a school posts a 3.86 GPA median and a 3.55 twenty-fifth percentile, it is disclosing that a quarter of its enrolled class came in at least three tenths below the middle, and those students almost all arrived with LSAT scores doing the work.
The asymmetry is structural rather than a matter of taste. The LSAT is a single scale administered identically to everyone, with a published percentile conversion; a GPA is a number generated by an institution with its own grading culture, its own major mix, and its own inflation history.
That makes the LSAT the cleaner signal and the GPA the more contextualizable one. A committee can read a 3.4 from a demanding program with an upward trend and adjust; a 158 reads as a 158 regardless of context.
So the reverse splitter's fix is almost always the retake, not the addendum. Our page on plateaus covers what the evidence says about actually moving the score rather than repeating the attempt.
Most schools use some form of index: a weighted formula combining LSAT and CAS GPA into a single sortable value, used to triage files rather than to decide them. The weights differ by school and are generally not published.
What is publicly observable is the outcome of that weighting, which is why the 509 percentiles are more useful than any leaked formula. If a school's 25th percentile LSAT is high and its 25th percentile GPA is low, it is telling you which number it protects.
Compare George Mason and GW on exactly this. Both post a 3.55 GPA twenty-fifth percentile, but Mason's LSAT twenty-fifth is 162 against a 169 median, a seven-point spread, while GW's is 162 against a 168 median, a six-point spread. Both protect the LSAT; neither protects the GPA nearly as hard.
This is the method, run by hand, that any odds calculator is approximating. It takes about ten minutes per school and it is more honest than a percentage from a model you cannot inspect.
Step one, pull the school's current 509 report and write down six numbers: LSAT 25th, 50th, 75th and GPA 25th, 50th, 75th. Use the school's own PDF, not an aggregator, because aggregators lag and occasionally transcribe wrong.
Step two, place each of your numbers in one of four zones: above the 75th, between the 50th and 75th, between the 25th and 50th, or below the 25th. Step three, read the pair against the grid below. Step four, adjust for timing, since the same pair submitted in October and in February are not the same application.
No school publishes admit rates by score band, so no honest page can give you a percentage. What follows is a directional reading built strictly from percentile position.
Your positionDirectional readWhat to doBoth above the 75thStrong. Money is the real question, not admissionApply early, negotiate aggressivelyBoth between the 50th and 75thCompetitiveClean file, early submission, expect some merit aidOne above 75th, one below 25thLive splitter or reverse splitterSplitter: addendum if there is a real story. Reverse splitter: retakeBoth between the 25th and 50thBelow the class midpoint but inside the rangeEverything early and flawless; aid unlikely at stickerOne at median, one below 25thUphill but realRetake if the low number is the LSATBoth below the 25thReachRetake, then rebuild the list
Percentile anchors come from each school's 509 report; the readings are our framework, not published admit rates. Treat anyone quoting you a precise percentage for a school that publishes none with suspicion.
Merit aid tracks the same percentile position that drives admission, because schools pay for the medians you deliver them. This is the part of the numbers game applicants systematically underweight.
The disclosed grant data shows how differently schools deploy that money. Georgetown's median grant was $35,000 to 61 percent of students, GW's $28,000 to 77 percent, and American's $40,000 to 59 percent, all from the 2025 509 reports covering the 2024-25 aid year.
SchoolMedian grantShare receivingGeorgetown$35,00061%GW$28,00077%American$40,00059%George Mason$21,000 full-time84%NYU$20,00066%
Source: each school's 2025 ABA 509 report, grant data reflecting the 2024-25 aid year.
Three tenths of an LSAT band can be worth more in aid than any other single change you can make in a cycle. That is the argument for treating the score as a financial instrument rather than an admissions hurdle, and the mechanics of converting it into dollars are in our negotiation playbook.
Use two tests, both cheap to run. First, is your practice-test average meaningfully above your official score? If yes, the gap is executional and a retake usually captures it.
Second, where does the gain land on the percentile curve? Four points from 158 to 162 moves you roughly fourteen percentile points and changes which tier you are shopping in; four points from 173 to 177 moves you about one and changes very little.
The base rate for calibration: the average second-attempt gain was 2.39 points in 2024-25, and 49.1 percent of test takers now test more than once (LSAC TR 26-01, https://www.lsac.org/sites/default/files/research/TR-26-01.pdf). Plan around two to three points, not around the outlier stories.
Timing is the third number, and it is the only one that is free. The 2026 cycle drew 82,189 applicants, up 8.1 percent year over year (LSAC Five Year Volume Comparison, July 22, 2026), and those applicants competed for seats and scholarship dollars that were committed progressively across the cycle.
Whether an early application carries a measurable admissions boost is genuinely disputed, and no school publishes admit rates by submission date. What is not disputed is that money and seats are finite and get spent.
So run the odds grid, then apply the correction: your read holds if you submit in the first third of the cycle, and degrades from there. The full sequencing sits in our admissions pillar.
Every figure on this page traces to an ABA Standard 509 Information Report, the disclosure each accredited school files annually in December. It is two pages, it is free, and it is the only source that is simultaneously current, standardized, and unspun.
Page one carries the admissions block: completed applications, offers, matriculants, and the LSAT and GPA percentiles for the entering class. Page two carries tuition, living expenses, and the grants table, including what percentage of students received aid and the 25th, median, and 75th percentile award.
Aggregator sites are convenient and they lag. During the window between December, when new 509s post, and the following spring, a large fraction of the internet is quoting last year's medians as this year's, which is exactly the season most applicants are building their lists.
First, the 25th percentile, which is the real floor and the single most useful number for anyone below a median. Rankings compress this away entirely; two schools with identical medians can have 25th percentiles a full band apart, and that difference is the whole question for a splitter.
Second, the offers-to-seats ratio, which tells you how much yield pressure a school is under and therefore how live its waitlist is. GW made 2,644 offers to seat 612 students in fall 2025, an implied yield near 23 percent by our arithmetic, while Georgetown made 2,193 offers to seat 672, near 31 percent.
Third, the grant distribution, which tells you whether a school buys medians with money and how deeply. A school awarding grants to 84 percent of students is running a different pricing strategy than one awarding to 59 percent, and you negotiate differently against each.
Treating the median as a cutoff. Half of every enrolled class scored below it, by definition, and those students were admitted deliberately rather than by oversight.
Treating the acceptance rate as your odds. It is an average across every applicant including those far above and far below the medians, so it describes the pool rather than you. A 27 percent school is not offering you 27 percent odds at either tail of its distribution.
Averaging your two numbers into a single mental score. Schools do weight them together, but they also report them separately to the ABA, which means a balanced pair and a split pair with the same average are not interchangeable in the way an average implies.
Finally, comparing schools on medians while ignoring cost. A school with a median one point lower and a median grant $12,000 higher is not a step down; it is a different price for a similar outcome, which is a decision rather than a compromise.
A 169 LSAT with a 3.42 GPA. At GW the LSAT sits above the 168 median and the GPA sits below the 3.55 twenty-fifth percentile, which is the textbook splitter shape at that school.
The read: live candidate, with the LSAT doing the work. The moves are a one-page factual addendum if the GPA has a documentable story, an early submission, and a scholarship expectation anchored on the LSAT rather than apologized for.
The same pair at Georgetown reads differently. The 169 falls between Georgetown's 166 twenty-fifth and 171 median, so it no longer carries the file, and the 3.42 is far under a 3.75 floor, which makes this a reach rather than a splitter play.
A 161 LSAT with a 3.94 GPA. At GW the GPA clears the 3.86 median comfortably while the LSAT sits below the 162 twenty-fifth percentile, which is the hardest common position to fix with words.
The read: the GPA is genuinely valuable and the LSAT is capping the file. A four-point gain to 165 moves this applicant from below GW's floor to inside its middle half, and on the percentile curve that gain is worth roughly thirteen points of percentile, since a 161 and a 165 sit at about 75 and 86.5 percent respectively.
That is the retake case in one paragraph, and it is why we treat the reverse splitter's cycle timing as flexible. Waiting one administration to test again usually beats applying now into a position the file cannot argue its way out of.
A 172 LSAT with a 3.95 GPA at GW, where the 75th percentiles are 170 and 3.93. Both numbers clear the top quartile, which changes the question from whether to how much.
At this position the application's job is to avoid unforced errors and to create leverage. That means an early, complete file, a clean statement, and applications to peer schools specifically so that competing offers exist when the aid conversation starts.
An applicant here who applies binding early decision has traded the most valuable asset in their file for a signal the school did not need. That is the most expensive common mistake made by strong candidates.
They cannot tell you whether you will do well in law school, since first-year performance depends heavily on skills the LSAT only partially proxies. They cannot tell you whether a given school's culture, market, or cost structure fits your actual plan.
And they cannot tell you the one thing every applicant most wants, which is a probability. No school publishes admit rates by score band, and every calculator that offers you one is modeling from self-reported data with unknown selection bias.
What the numbers do is bound the problem. They tell you which schools are worth an application, where you sit relative to the class you would join, roughly what money to expect, and whether another test administration is the highest-value use of the next three months.
The belief that schools reject overqualified applicants to protect their yield is one of the most confidently repeated claims in admissions, and it is close to unfalsifiable from published data.
What is documented is the structural incentive: schools care about yield, and GW's 2,644 offers for 612 seats implies a yield near 23 percent by our arithmetic. What is not documented is any school publishing admit or deny rates by score band, which is what a yield-protection claim would require to be tested.
Our position is that the honest answer is unknown, and that applicants who suspect it should respond by demonstrating genuine interest rather than by lowering their targets. Concrete engagement costs nothing and addresses the concern directly.
Run the odds engine on every school on your list and lay the readings side by side. The pattern that emerges is usually more informative than any single school's read.
A list where every school reads reach means the list is wrong, not that you are unqualified. A list where every school reads likely means you are underselling your numbers and should add schools upward.
The healthy shape is roughly a third reaches, a third targets, and a third likelies, with at least two schools where you sit above both 75th percentiles so that real money is in play. That last group is what funds the rest of the decision.
Before deciding, model it. Take your target score, find its percentile, and rerun your position at each school on your list rather than reasoning about the number in the abstract.
A 163 moving to 167 crosses from roughly the 80th to the 90th percentile and changes GW from a reach to a target, since GW's 25th is 162 and its median 168. The same four points from 172 to 176 crosses from about the 97th to the 99th and changes almost nothing about which schools are viable.
That is the whole retake calculus in two sentences. The value of the retake is not the points, it is what the points do to your position at the specific schools you actually want.
If you are still in undergrad, the numbers on this page are partly under your control, and the leverage is larger than most students realize.
LSAC counts every attempt at a repeated course with no grade replacement, and excludes all coursework after your first bachelor's degree. That combination means credits earned now count fully while a post-graduate masters will not move your CAS GPA at all.
The practical implication for a current student: an additional strong semester before graduating is worth more to your law school application than a graduate degree afterward. Almost nobody sequences it that way, because the rule is counterintuitive and lives in a document few people read before their senior year.
Your LSAT and GPA do not do the same job inside a school's decision, and understanding the division explains most outcomes that look inconsistent.
The LSAT is a prediction and a comparison. It is administered identically to everyone, its percentile conversion is published, and it is the number most closely watched in the reporting that shapes a school's public position.
The GPA is a history. It reflects four years, an institution, a major, and a grading culture, all of which a reader can contextualize and none of which transfer cleanly across applicants.
That is why the same spread produces different outcomes for splitters and reverse splitters, and why the advice on this page is asymmetric. It is not a preference for test takers, it is a consequence of what each number can and cannot tell a reader.
Odds calculators are popular because the uncertainty is uncomfortable, and they trade on data that cannot support the precision they display.
They are built from self-reported outcomes on public forums, which is a sample selected for people who chose to report, skewed toward strong results and unusual cases. No school publishes admit rates by score band, so there is no authoritative dataset for any calculator to be calibrated against.
Use them for shape if you like them, and do not let a number with two decimal places substitute for the percentile work on this page. A calculator that tells you 34.7 percent is expressing a confidence that its inputs do not license.
The method here is deliberately coarser and more honest. It tells you which zone you are in, what that zone typically means, and what action it implies, which is everything a percentage would have given you minus the false precision.
For the T14, medians ran 170 to 174 for fall 2025 entering classes, so a competitive target is 170 and above, with 173+ for the very top. Strong regional schools sit in the 163 to 169 band and most ABA schools in the 155 to 162 band.
It is just below the national median of about 3.65 and works at most schools with a strong LSAT. It sits between the 25th and the median at American, just under the 25th at GW and George Mason, and well below the 25th at Georgetown and NYU.
An applicant whose LSAT is at or above a school's median while their GPA is at or below its 25th percentile. Splitters get admitted regularly because schools report both medians and the LSAT is the harder one to acquire.
Less often at the same spread, because the LSAT is a uniform national scale while a GPA can be contextualized by program and grading culture. For most reverse splitters the retake outperforms every other available move.
It depends entirely where you start. Five points is worth roughly 17.9 percentile points at a 150 and 1.0 percentile point at a 175, so the same gain is transformative low on the scale and marginal at the top.
Written by Ali, Georgetown Law, founder of Lovare Institut.
July 30, 2026
July 30, 2026