LSAT Logical Reasoning Guide

Since the August 2024 change, the LSAT has three scored sections: two Logical Reasoning and one Reading Comprehension, each 35 minutes. That means LR is roughly two-thirds of your scored questions. LSAC doesn't publish a fixed question count or an official breakdown by question type, but each LR section typically runs about 25 questions—about 1 minute 24 seconds apiece.

Every question has the same parts: a short stimulus (an argument, a set of facts, or a two-person exchange), a question stem telling you what to do, and five answer choices. LSAC describes the section as testing your ability to "analyze, critically evaluate, and complete arguments." No outside knowledge is required, and you shouldn't bring any in.

There are more than a dozen question types, but they fall into four families. Identify the family from the stem before you read the stimulus—it tells you what to look for.

The Linden County library system should stop buying printed encyclopedias. Last year, patrons used the library's online encyclopedia database more than 20,000 times, while the printed encyclopedias were taken off the shelf fewer than 300 times.

The argument depends on which one of the following assumptions?

1. The online database costs less than the printed encyclopedias.
2. The online database provides substantially the same information as the printed encyclopedias.
3. No patron used both the printed encyclopedias and the online database.
4. Most library patrons have computers at home.
5. The printed encyclopedias contain errors that the online database has corrected.

Answer: B. Conclusion: stop buying print. Evidence: people use the database far more. The gap: usage numbers only matter if the database can replace what print offers. Use the negation test—flip the answer and see if the argument collapses. "The database does not provide substantially the same information" means dropping print loses content, and the usage comparison no longer justifies the recommendation.

Why A is tempting but wrong: cost feels relevant, but the argument never compares prices, and the database stays either way. Negate A (the database costs more) and the argument still stands. C and E are far stronger than the argument needs, and D is irrelevant: the usage numbers already show that patrons can reach the database.

A survey found that people who drink green tea every day report fewer heart problems than people who don't. So drinking green tea every day protects against heart problems.

The reasoning is most vulnerable to criticism on the grounds that it…

Before you look at the choices, predict: this jumps from a correlation to a cause. The right answer will say something like "fails to rule out that some other factor associated with drinking green tea—such as diet or exercise—accounts for the difference." A tempting wrong answer might say the argument "relies on a sample that is too small"—but nothing in the stimulus tells you the sample size, so that criticism has no support.

Predicting the flaw before reading the choices is the fastest way to avoid choices that sound bad but don't describe what this argument actually did.

Many assumption, inference, and parallel questions turn on "if–then" statements. Four rules cover most of them:

• Sufficient → necessary. "If you're a member, you get a discount" means member → discount. Being a member guarantees the discount; getting a discount doesn't prove membership.
• Contrapositive. Reverse and negate: no discount → not a member. This is always valid. Reversing without negating (discount → member) is the classic error.
• "Only if" and "requires" introduce the necessary condition. "You can vote only if you're registered" = vote → registered.
• "Unless" means "if not." "No one may join the team unless they pass the physical" = not pass → not join (equivalently, join → passed).

Also watch quantifiers. "Most A are B" plus "most A are C" guarantees that some B are C. But "most A are B" plus "most B are C" guarantees nothing about A and C—a favorite parallel-flaw trap.

• Correlation treated as causation, or ignoring that the cause could run the other way.
• Unrepresentative sample—for example, drawing conclusions about everyone from people who chose to respond.
• Necessary vs. sufficient confusion—treating a requirement as a guarantee, or vice versa.
• Attacking the source instead of the claim.
• Shifting meaning of a key word between premise and conclusion.
• Part vs. whole—assuming what's true of each member is true of the group, or the reverse.
• Percentages vs. numbers—a share can rise while the total falls, and vice versa.
• Absence of evidence treated as proof that something is false.
• Not necessary means no effect—concluding a factor doesn't matter because the outcome sometimes happens without it.

• Too strong: "all," "never," "only" in an inference answer when the stimulus supports "some."
• Out of scope: the choice discusses something the argument never touches, however reasonable it sounds.
• Reversed: flips a conditional or a causal direction.
• Restated premise: in assumption questions, a choice that repeats what's already stated isn't an assumption.
• Wrong direction: a strengthener offered on a weaken question, or the reverse. Check the stem again before you commit.
• Intermediate conclusion: on main-conclusion questions, a claim that follows "so" or "therefore" may just be a stepping stone to the real point.

At about 84 seconds per question, you don't need to rush—you need to avoid sinking three minutes into one question. Work in order, answer and flag anything you're unsure of, and move on. When you finish the section, return to flagged questions with the time you banked. Because wrong answers aren't penalized, never leave a question blank; pick your best guess before time expires.

1. Week 1 — Structure. Main conclusion, role, and method questions. For every stimulus you read this week, write the conclusion in your own words before looking at the stem.
2. Week 2 — Assumption family. Necessary and sufficient assumption, then strengthen, weaken, and flaw. Practice the negation test and learn the flaw list above.
3. Week 3 — Inference family and conditional logic. Must be true, most strongly supported, principle, and parallel questions. Diagram conditionals until the contrapositive is automatic.
4. Week 4 — Timed sections. Two or three timed 35-minute LR sections, each followed by blind review. Keep an error log: question type, why you picked the wrong answer, and what the right answer did.

Then test yourself with the two timed LR sections in our free LSAT practice test and a free official PrepTest on LawHub. For the other scored section, see the Reading Comprehension guide; for dates, fees, and format, see the LSAT overview. If you want a book, our best LSAT prep books list flags the LR-specific options.