The 2 Numbers That Read First
One is frozen. One is live. Act accordingly.
Most applicants have a relationship with their numbers instead of a plan for them. The 3.9s treat the GPA like a trophy that should be self-explanatory. The 3.3s treat it like a scar they hope nobody mentions. People waiting on an LSAT score treat the whole subject like weather. Shame, pride, weather: none of these is a strategy, and all of them get in the way of the only question that pays: what do these 2 numbers actually do once they’re in the file?
You learned in chapter 1 that they get read first and set the question the rest of your file answers. This chapter goes further: what the scale looks like in real data, which of your 2 numbers can still move and by how much, what happens when your numbers disagree with each other, and what your particular pair means for how you’ll build a list in chapter 11. Evidence first. Argument later. It’s how you’ll practice law, so it may as well be how you get into it.
The scale you’re actually on
Numbers first, real ones, dated. Across the 196 law schools with published medians for the class that entered in fall 2024, the median LSAT ranges from 139 to 175. The middle school on that list sits at 159. Up top, 22 schools hold medians of 170 or higher and another 26 sit between 165 and 169. In the broad middle, 43 schools land between 160 and 164 and 56 between 155 and 159, and 49 schools have medians below 155. Median GPAs run from 3.09 to a flat 4.0, which yes, means at least 1 school’s typical admit carries a perfect transcript. (All of this is computed from the schools’ own ABA disclosures, which they’re required to publish annually. When you read this, pull the current year’s; the shape moves slowly, the specific numbers move every cycle.)
2 things to take from the real scale before we do anything clever with it.
First: wherever your numbers land, schools exist on both sides of them. The question is never the forum question, “am I good enough?” It’s the venue question: at which schools is my evidence strong, neutral, or a problem? Same numbers, 3 different roles, depending entirely on where you file.
Second, and almost nobody metabolizes this one: a quarter of every entering class sits at or below the school’s published 25th percentile. That’s not a loophole, it’s arithmetic; it’s what a percentile is. The gap between a school’s 25th and 75th LSAT percentile averages about 6 points in that same 2024 data. Schools admit below their medians all season, on purpose, for reasons (and at prices) we’ll get to. So a median is not a wall. It’s the table stakes of the close call.
The frozen number
Now the uncomfortable audit. Your GPA is mostly sunk, and I’d rather show you than assert it.
Say you’re carrying a 3.4 through 6 semesters and you find religion: straight 4.0s from here to graduation. Quality points: 3.4 times 6 is 20.4, plus 4.0 times 2 is 8, divided by 8 semesters. You graduate at 3.55. 2 perfect semesters, every class, no slips, buys you 0.15. If you’re a senior reading this in the fall, the live fraction is smaller still. This is why I have no chapter for you on “GPA repair.” The repair window is the years you already spent, and anyone selling you otherwise is selling.
3 things about the frozen number are still live, though, and they’re worth knowing exactly.
The trend is readable. A 3.4 built from a 2.9 freshman year and a 3.8 junior year is a different document than a 3.4 built from steady 3.4s, and readers do read transcripts, not just summaries, when you’re in the hard pile. You can’t change the number; the shape already argues for you or doesn’t.
The computation is checkable. LSAC calculates its own undergraduate GPA from every undergraduate grade you’ve ever earned, including the retakes your college forgave and that summer course at the community college. You did this homework in chapter 3; if you skipped it, it’s still 30 minutes, and finding a surprise now beats finding it in a reader’s hands.
And graduate grades don’t enter it. LSAC’s number is undergrad only. A master’s program can say things about you (Part 3 material), but it cannot retroactively touch the GPA that feeds the medians. People spend 2 years and $60,000 learning that sentence the slow way.
Everything else about the GPA goes through chapter 10, where you’ll learn when a dip earns an addendum and when explaining does more damage than the dip.
The live number
The LSAT is the opposite animal: fully movable until the moment you file, and governed by rules that are friendlier to you than most applicants ever find out.
Here’s the part worth money. Law schools receive your full testing history from the reportable period, so yes, they see every score. But the number a school must report to the ABA for its entering class, the number that feeds its medians, is your highest. That asymmetry runs the whole retake market: a 161 sitting next to an old 156 costs the school’s published profile nothing, because only the 161 counts where the school is counted. Most schools say plainly that they consider the highest score, and the incentive math explains why you can believe them. A bad earlier score isn’t a stain you carry. It’s a draft.
The practical consequences, in order. A delivered score usually beats a cancelled one, because it’s data and it’s floor. A retake with a real preparation change behind it is one of the few moves in this entire process with asymmetric upside: capped downside at most schools, uncapped benefit. And the current rules give you 5 takes within the reportable period and 7 lifetime, with no per-year cap, so scarcity is real but not the emergency the forums perform. (Limits drift; confirm the current ones at lsac.org before you plan around mine.)
What this chapter will not do is teach you to raise the number. That’s a whole other book, and I wrote it: Anyone Can Beat the LSAT, which exists because 2 or 3 more points routinely outvalue everything in Part 4 of this one. What this book owns is the decisions around the score, and they’re worth their own chapter, which is next. For now, lock in the asymmetry: of your 2 numbers, exactly 1 still has a steering wheel attached. Spend accordingly, which for most readers means: the LSAT gets your hours until it’s done, and the essays get your hours after, and people who run that order backward pay for it twice.
When your numbers disagree
Now the case the internet argues about endlessly and explains never: the split.
Run the twins. Theo carries a 3.85 and a 158. Bria carries a 3.45 and a 168. Both apply to the same school, call it Marbury Law, which published medians of 3.70 and 161 last cycle. A points-based view says they’re roughly equivalent: each is comfortably above 1 median and meaningfully below the other. Mirror images. The room does not see mirror images.
Sit in the reader’s chair with the class-building ledger open. It’s January. Theo’s admit nudges the GPA median up, which is nice, and hangs a 158 below the LSAT median, which has to be paid for: some other admit’s 164 now exists to offset Theo instead of lifting the published number. Bria is the reverse: her 168 is lift, her 3.45 is drag. Asking which one gets the offer is the wrong question. The real question, the one that flips outcomes: which median is this school defending this year? If Marbury’s 161 is fragile (last year it slipped a point; the dean noticed), Bria’s 168 reads like aspirin. If its GPA median is the contested number, Theo’s the aspirin. Same 2 files, opposite verdicts, at 2 schools that look identical from the outside, or at the same school in 2 different years. When a splitter tells you their results “made no sense,” the results made perfect sense. The sense lives in data they never pulled.
So if your numbers disagree, your whole strategy is venue research plus volume. You can read a school’s defended number in its published history, and since “go read disclosures” is the kind of advice that sounds easy until you’re staring at one, let’s read one together. Here’s Marbury Law across 3 invented but realistic years, the 3 numbers that matter for each: LSAT 25th/median, GPA 25th/median.
2 cycles ago: LSAT 158 / 162. GPA 3.48 / 3.71 Last cycle: LSAT 157 / 161. GPA 3.52 / 3.74 This cycle: LSAT 157 / 161. GPA 3.55 / 3.76
Read it like a lawyer. The LSAT median slipped a point and then held, and the LSAT 25th sagged with it. So Marbury is straining to hold 161. So LSAT points are what it’s short on. So Bria’s 168 isn’t just welcome, it’s budget relief, and she should expect an admit with money attached (chapter 13 will show her how to ask for more of it). Meanwhile both GPA numbers climbed 2 years running: GPA is the median Marbury can afford to be picky about, which means Theo’s 3.85 buys him less here than he’d hope and his 158 costs more. Same school, same webpage, and 10 minutes of reading just told the twins opposite things, both true. That’s venue research. Unglamorous, 10 minutes a school, and it’s the difference between a splitter cycle that “made no sense” and one that was legible before the first fee was paid.
The rest of the splitter playbook follows from it. Apply wider than your friends with tidy profiles. Expect variance instead of being insulted by it. And remember the price tag: splitters who clear a school’s tight median on the number it’s defending get courted. Chapter 13 teaches you to collect on that.
But Bria’s mirror image deserves his own page, because the internet treats reverse splitters as a punchline and the truth is just different math. So: Theo’s playbook, 3.85 and a 158.
Start with what Theo has that Bria doesn’t: a number that took 4 years of showing up to build, and that nobody can question as a fluke. GPA medians get defended too. Plenty of schools sit where Marbury doesn’t: their LSAT median is comfortable and their GPA median is the number straining, often because they’re chasing the rankings inputs that reward it or repairing a weak year. Theo’s disclosure read is the same skill pointed the other way: he’s hunting for schools where the GPA 25th sags or the GPA median slipped, because those schools are quietly shopping for exactly his shape. They exist every cycle. They’re just never the same schools the forums are shouting about. That’s why he has to read the filings instead of the threads.
Then the harder honesty, and Theo needs it straight: his 158 is the live number. Bria can’t retroactively fix a 3.45; Theo can absolutely take the LSAT again. So before he builds a single list around 158, chapter 5’s question lands on him with full weight: is the score done, by the actual conditions, or is it just declared done because studying was miserable? A 4-point LSAT move would do more for Theo’s options than any venue strategy, because it un-splits him. The reverse splitter’s playbook, in order. First, decide the score honestly. Then hunt the GPA-defending venues. Then apply early; his profile reads as low-risk, and committees like resolving low-risk files while budgets are full. Then negotiate like the evidence he is. A 3.85 above a school’s GPA median is median protection wearing a deposit.
One more case, because some of you are reading this chapter with 1 number and a blank: no LSAT yet. Your position table still works; you just run it twice. Once with your practice-test average, honestly stated, to see the shape of your likely options. Once with that average plus 3, to see what’s on the other side of the prep you haven’t finished. The gap between those 2 tables is the dollar value of taking chapter 5 seriously, written in school names instead of abstractions. People who run both tables stop needing to be nagged about prep. The second table does the nagging.
Do this now: 25 minutes
The position table is the chapter’s assignment, and it has a part 2 tonight. Part 1, from before: 6 schools, current numbers, your position at each in 1 word. Part 2: for the 2 schools that interest you most, pull 3 years of their disclosures and write 1 sentence each, Marbury-style: which number is this school defending, and what does that mean for someone shaped like me? If you can write those 2 sentences, you’ve learned the skill most paid consultants are selling. Date the page. Chapter 11 builds on it directly.
The conditional verdicts, plainly. Above both medians: strong position, mind chapter 1’s yield-protection note, and start thinking of admission as the opening bid, not the prize. Split: no single answer exists, unless you do the venue research, in which case patterns emerge fast. Below both: not a no, but a “no unless,” and the unless has 3 parts. Unless the school’s 25th percentiles say people like you enroll there every year. Unless something in your file (Part 3 and 4’s job) answers the question a seat-spending reader has to win in committee. And unless you’ve priced the offer realistically, because below-both-medians admits get the worst scholarship offers. That last clause is chapters 11 and 13. It is not a detail.
What about everything that isn’t a number?
Here’s where I’m supposed to either tell you softs are decisive (hype) or that nothing matters but the index (doom). The room says: softs are evidence, not points. Work experience, research, service, the debate trophy, the dispatch job: none of it adds 0.1 to anything. A reader triaging at chapter 1 speeds has no rubric slot for “impressive, vaguely.” What softs do, in the hard pile where you live, is decide whether your theory of the case is provable. A claim with 4 years of dispatch logs behind it beats a claim with adjectives behind it, every read. That’s why this book spends all of Part 3 turning your raw material into an argument and exactly 0 pages ranking extracurriculars. The inventory you’ll build in chapter 7 is where every soft you own gets weighed, the only way the room weighs them: as exhibits.
Try it: the position table
20 minutes, into the case file. List 6 schools you’re currently imagining, honest mix. For each, pull the current published medians and 25th percentiles (the school’s ABA disclosure page has them; so does LSAC’s school directory) and write your position in 1 word per school: above (both), split (and which way), or below (both). Date the page.
You now hold a map of your own variance: at which schools your numbers do the arguing, at which they’ll need the file’s help, and at which the file is carrying everything.
Then part 2, 10 more minutes, for the 2 schools that interest you most: pull 3 years of their disclosures and write 1 sentence each, Marbury-style: which number is this school defending, and what does that mean for someone shaped like me? If you can write those 2 sentences, you’ve learned the skill most paid consultants are selling. No decisions yet. Chapter 11 turns this table into a list and a budget; today it’s enough that nobody can hand you a mood about your chances again. You’ve seen your actual positions, plural. That’s what chances are.
One number frozen, one live, both legible to anyone who looks. That’s your evidence locked. Next: the handful of decisions around the live number (when to take it, when to take it again, when to stop) that quietly move more money than any document you’ll write.
