“"The practice questions closely matched the official Logical Reasoning style and helped me become much more confident before test day."”
Jessica Miller
Aspiring Law Student
Since the LSAT's August 2024 redesign, every test includes two scored Logical Reasoning sections, each holding 24 to 26 questions and running 35 minutes. That's roughly a question every 84 seconds if you work straight through — except you won't, because some stimulus-and-five-choice arguments take fifteen seconds to solve and others eat three minutes on their own. Together the two LR sections make up close to two-thirds of your scored questions, somewhere around 50 of the roughly 77 questions that count toward your 120–180 scale. That weighting is the single biggest reason a weak LR section drags a whole score down even when Reading Comprehension goes well.
Format-wise, every question follows the same shell: a short stimulus (usually two to five sentences describing an argument or situation), a question stem that tells you what to do with it, and five answer choices. There's no separate instruction page reminding you which task you're doing — you have to read the stem correctly, every time, under time pressure. That's the actual skill being tested, more than raw logic.
You'll also face one unscored experimental section mixed in among the scored ones, indistinguishable in the moment — it could be a third Logical Reasoning section or a second Reading Comprehension section. You won't know which section is the throwaway, so pacing yourself as if every section counts is the only sane strategy.
LSAC doesn't publish an official percentage-by-type table, and outside analysts who track released tests don't agree down to the decimal — their counts shift depending on which test administrations they sampled and how they classify borderline questions. What's consistent across every serious analysis is the ranking: a handful of question types dominate every test, and the rest show up occasionally.
Question Type | Approximate Share | What It's Testing |
|---|---|---|
Flaw / Error in Reasoning | 14–16% | Naming the specific logical error in an argument |
Inference / Must Be True | 14–15% | Identifying what's provable from the stimulus, not what sounds plausible |
Necessary Assumption | 9–11% | Finding the unstated premise the argument requires to hold |
Strengthen | 7–14% | Picking the choice that most supports the conclusion |
Weaken | 7–9% | Picking the choice that most undermines the conclusion |
Parallel Reasoning / Parallel Flaw | 4–8% | Matching argument structure across an unrelated topic |
Method of Reasoning | 6–7% | Describing the argumentative technique used |
Main Point | 6–7% | Isolating the conclusion, not a supporting premise |
Resolve the Paradox | 5–6% | Explaining how two apparently conflicting facts can both be true |
Sufficient Assumption / Justify | 5–6% | Finding the premise that would make the argument airtight |
Point at Issue / Point of Agreement | 2% | Identifying exactly what two speakers do or don't disagree about |
Evaluate the Argument | 1–2% | Naming the question whose answer would determine the argument's strength |
Assumption, Flaw, and Inference alone typically account for about 40% of a given Logical Reasoning section, and adding Weaken and Strengthen pushes that past 60%. That's the practical takeaway: five question types decide most of your score, so uneven prep time across all twelve-plus types is a mistake even experienced test-takers make.
Conditional logic shows up far more than people expect, and it's rarely labeled as such. Sufficient/necessary reasoning ("if A, then B") drives correct answers on Necessary Assumption, Sufficient Assumption, and a chunk of Flaw and Inference questions — but it rarely appears as a clean textbook conditional. It's buried in phrases like "only if," "unless," "no ... without," and "the only way." Candidates who never formalized this notation hit a wall on medium-to-hard questions that look like reading comprehension but are actually symbolic logic wearing a disguise.
Causal reasoning flaws get memorized as a category, then missed in practice. Everyone learns "correlation isn't causation" as a flaw pattern. Fewer people can reliably spot the four or five specific ways an LSAT argument commits it — reversed causation, an unconsidered third cause, coincidence dressed as causation, or treating correlation over time as proof. Recognizing the label isn't the same as catching it live in a 30-second read.
Necessary Assumption and Sufficient Assumption get confused constantly, because both use the word "assumption" in the stem and both feel abstract. The negation test resolves this reliably — negate the answer choice and see whether the argument falls apart — but most self-studiers skip building that habit early, then rely on gut feel under time pressure, where it fails them.
Wrong answers are wrong for structural reasons, not vibes. Extreme language, answers outside the scope of the stimulus, answers that address a different part of the argument than what's asked, and answers that are half right and half wrong are the four most common wrong-answer traps. Candidates who treat wrong-answer review as "well, that one was less right than the correct one" instead of naming the specific trap plateau far earlier than candidates who do.
Logical Reasoning rewards a slower kind of practice than most people default to. Because the section is entirely question-type-driven rather than passage-driven, the highest-value practice isn't timed sets — it's untimed, type-isolated drilling until accuracy on a given type is consistently high, then adding the clock.
A workflow that actually moves scores:
Drill one question type at a time, untimed, until you can articulate why each wrong answer is wrong — not just that it "felt off." If you can't explain the flaw in a wrong choice in one sentence, you got the question right by luck, not by method.
Blind review before checking explanations. Re-read a question you got wrong (or guessed on) with no time pressure and no answer key. If you can now solve it, the error was pacing, not comprehension — a very different fix.
Track errors by type, not by test. A log that says "missed 6 questions on Test 3" tells you nothing. A log that says "missed 4 Necessary Assumption questions this week, all involving comparison flaws" tells you exactly what to drill next.
Time full sections only after per-type accuracy stabilizes. Adding a clock to a skill you haven't built yet just trains you to panic faster.
Use full, timed practice tests sparingly and strategically — as calibration checkpoints roughly every two to three weeks, not as the default study activity. Most of the actual score gain happens in slow, type-isolated review, not in racking up timed-section reps.Take the Law Enforcement Analyst – Foundational (LEAF) practice test for free.
Skipping the negation test on Necessary Assumption questions and relying on which answer "sounds necessary," which is exactly the intuition the wrong answers are built to exploit.
Confusing "strongly supported" with "guaranteed true" on Inference questions — picking an answer that's plausible rather than one the stimulus actually proves.
Running out of time on the last four to five questions of a section because early questions got more time than they needed. LR punishes uneven pacing more than raw difficulty.
Treating every wrong answer choice as equally wrong, instead of noticing that on hard questions, LSAC usually plants one seriously tempting distractor designed to catch exactly the shortcut you took.
Studying almost exclusively from one prep company's materials. Question-writing style varies slightly by source; candidates who only ever see one company's phrasing sometimes struggle with real LSAT stem wording on test day.
Not reviewing correct answers, only wrong ones. Getting a question right for the wrong reason is a trap that resurfaces on a harder version of the same question type later.
Beginner (little to no formal logic or standardized test background): Expect 8–12 weeks focused mostly on Logical Reasoning before layering in full sections. The first two to three weeks should go almost entirely to conditional logic notation and the five highest-frequency question types — trying to time yourself before that foundation is set usually backfires and just trains bad habits at speed.
Has related experience (strong reading comprehension background, debate, philosophy, or a prior standardized test with heavy verbal reasoning): 6–8 weeks is typically enough to convert existing argument-analysis instincts into LSAT-specific pattern recognition, with more time spent on timing and the trickier assumption/flaw subtypes than on basics.
Retaking: Timeline depends entirely on what the score breakdown from the previous attempt shows. If LR was the weak section, isolate exactly which question types were missed most before doing anything else — retaking without that diagnosis just repeats the same errors at the same pace. Four to six weeks of targeted drilling on two or three specific weak types, followed by two or three full timed sections, is usually more productive than another full-length repeat of general prep.
Logic Games (Analytical Reasoning) was permanently removed from the LSAT before August 2024, which shifted more overall test weight onto Logical Reasoning and Reading Comprehension rather than changing LR's internal structure. Starting with the 2026 testing year, the LSAT also added an Argumentative Writing task taken separately online — this doesn't affect the Logical Reasoning section itself, but it does affect overall test-day planning and timeline for candidates targeting a 2026 or later test date. The core LR question types and their relative frequency have stayed fairly stable across recent years; what shifts test-to-test is which specific flaw patterns and topics get more emphasis, not the underlying skill set being tested.
Last updated on Oct, 7 2026