Logical Reasoning makes up two of the three scored sections on the LSAT, and about two-thirds of all the questions that count toward your score. So it is worth knowing what it can actually ask you.
Most guides answer that with an alphabetical list of a dozen or so type names. That list is not much use in a timed section, because the names don't tell you what to do. Knowing that a question is called "Necessary Assumption" is only valuable if you know that the label means: find the thing this argument silently needs, and expect the right answer to sound underwhelming.
So here they are grouped by task instead. LSAC never publishes an official list, and every prep company draws the lines somewhere slightly different; we track nineteen types, and they fall into four families. Learn the family and you know the job before you have read a word of the answer choices.
Four things a question can ask you to do
- Structural analysis — describe what the argument is doing, without judging whether it's any good.
- Argument evaluation — find what would make the argument stronger, weaker, or testable.
- Assumption analysis — find what the argument is taking for granted.
- Deductive inference — work out what does or doesn't follow from what's on the page.
There's a useful split hiding in that list. The first three families all require you to find an argument: a conclusion, some evidence, and a gap between them. The fourth family often has no argument at all, just a set of facts, and your job is to stay strictly inside them. Noticing which mode you're in is half the battle, because the reading habits that earn points in one will lose them in the other.
Everything below assumes you're working the question in the order described in analyze, predict, eliminate. The type tells you what to predict.
Structural analysis: what is this argument doing?
These questions ask you to describe an argument's anatomy. Crucially, most of them don't care whether the argument is convincing. They care what it is made of.
Flaw in the Reasoning
Names the gap in the abstract. "Which one of the following most accurately describes a flaw in the argument's reasoning?"
Predict by saying the gap in plain language, then look for the choice that says the same thing in the LSAT's formal vocabulary. The trap is a choice that describes a real, recognizable logical flaw that this particular argument didn't commit. Those choices are correct-sounding precisely because they're textbook.
Try this one:
Archaeologist: At every ancient Varnese site excavated so far, diggers have unearthed hundreds of intricately crafted clay pots, yet scarcely any wooden objects, and the handful of wooden items recovered are crude. The natural conclusion is that the Varnese were highly skilled potters but only rudimentary woodworkers.
Which one of the following most accurately describes a flaw in the argument's reasoning?
- It infers that the Varnese placed little value on the craft of woodworking solely from the absence of any direct evidence that they valued it at all.
- It overlooks the possibility that one kind of object survived far better than another, so the recovered finds may misrepresent the Varnese's true skills.
- It assumes that because crude woodworking is found alongside skilled pottery at these sites, the one must somehow have been the cause of the other.
- It relies on the word "crafted" in one sense in stating its premises and in a quite different sense in stating its conclusion about the Varnese.
- It draws a sweeping conclusion about an entire ancient people from the artifacts that happened to be unearthed at just a single one of their settlements.
Show the answer
B is correct. The evidence is about what survived in the ground; the conclusion is about what the Varnese could make. Wood decays and fired clay doesn't, so the scarcity and crudeness of the wooden finds may be a fact about preservation rather than a fact about skill. The argument treats what was dug up as a fair sample of what was originally produced.
A swaps the conclusion for a different one, since the archaeologist claims the Varnese lacked woodworking skill, not that they didn't value woodworking. C names a causal confusion the argument never makes; the two crafts are treated as separate measures, not as cause and effect. D alleges an equivocation, but "crafted" means the same thing throughout. E is the closest thing to a real second-place answer, and it fails on one word: the stimulus says "every ancient Varnese site excavated so far," not one site.
Notice that all four wrong answers name flaws that genuinely appear on the LSAT all the time. That's the type's signature difficulty. You cannot pick by asking "is this a real flaw?" You have to ask "is this this argument's flaw?"
Method of Reasoning (Argument)
Describes the argument's strategy, not its content. "The argument proceeds by…" Predict in structural terms before looking: it establishes a general rule and applies it to a case, or it rejects a hypothesis by pointing to a counterexample. Right answers are written at that level of abstraction; wrong ones usually describe a method the argument didn't use, or re-state the content rather than the method.
Method of Reasoning (Role of Statement) and Argument Part Identification
Point at one sentence and ask what job it does. "The claim that… plays which one of the following roles?"
Before reading the choices, decide two things: is the referenced statement a premise, the main conclusion, or an intermediate conclusion (something the argument supports and uses to support something else)? And is it something the author endorses, or a view the author is arguing against? Wrong answers on these questions frequently get the direction right but the level wrong, calling an intermediate conclusion the main one.
Parallel Reasoning (Valid) and Parallel Flaw
Ask you to find the answer whose logical structure matches the stimulus. The subject matter will always be different, and the resemblance in topic is bait.
Strip the stimulus to a skeleton before you read the choices, something like all A are B; this is an A; so this is a B, and match on that. For Parallel Flaw, the match has to include the error itself: the right answer commits the same mistake, not merely some mistake. These are among the most time-expensive questions on the section, and skipping one to bank ninety seconds for two others is often the correct play.
Argument evaluation: what would change it?
Here you're not describing the argument, you're acting on it. Every question in this family runs through the gap you found in step one.
Strengthen
"Which one of the following, if true, most strengthens the argument?" Predict something that closes the gap, usually by ruling out an alternative explanation or confirming that the evidence transfers to the conclusion's claim. Note that the right answer does not need to prove the conclusion. It only needs to help. A choice that merely restates a premise adds nothing and is a common wrong answer.
Weaken
"Which one of the following, if true, most weakens the argument?" Predict something that makes the gap worse: an alternative cause, a counterexample, a reason the evidence doesn't transfer. The single most common wrong answer attacks a claim adjacent to the conclusion rather than the conclusion itself, such as whether the plan is worth it when the conclusion was only about whether it works.
Evaluate the Argument
"Which one of the following would be most useful to know in evaluating the argument?" Rarer, and a Weaken question in disguise. The right answer is a question whose two possible answers pull in opposite directions: one answer would strengthen the argument, the other would damage it. If a choice's answer wouldn't change your assessment either way, it's out.
Paradox (Resolve, Reconcile, Explain)
"Which one of the following, if true, most helps to resolve the apparent discrepancy?" The odd one out in this family: there's no argument, just two facts that sit uncomfortably together. State the tension precisely, as in X should have led to Y, but Z happened instead, then look for the choice that lets both facts be true at once. Wrong answers typically make the puzzle worse, or explain only one side of it.
Assumption analysis: what is it taking for granted?
The family that trips up the most people, because two of its types sound identical and behave completely differently.
Necessary Assumption
"Which one of the following is an assumption required by the argument?" You're looking for something the argument needs in order to work, not something that would make it airtight.
The test is negation: if you flip the choice to its opposite and the argument collapses, it was required. If you flip it and the argument survives, it wasn't. Because these answers only have to be necessary, they're often strikingly modest, along the lines of "the two groups did not differ in some other relevant way" rather than anything sweeping. Students routinely reject the correct answer for being too weak, then pick a confident-sounding choice that the argument doesn't actually require.
If an argument concludes that a drug helps some patients, it does not require the assumption that the drug helps all of them. Negate "helps all" and the argument is untouched, so that assumption was never necessary. Any word like only, never, always, or must in a Necessary Assumption choice deserves suspicion.
Sufficient Assumption
"Which one of the following, if assumed, allows the conclusion to be properly drawn?" The mirror image. Here you want something enough to make the conclusion follow, and being too strong is perfectly fine.
These questions reward finding the exact missing link: something in the premises, something in the conclusion, and a bridge that doesn't exist yet. Predict the bridge before you look, as in the argument needs "everyone in group X also has property Y," and the matching answer is usually unmistakable.
Principle (Identify) and Principle (Apply)
Principle questions move between a general rule and a specific case. Identify questions give you the case and ask which broad principle it illustrates or justifies; Apply questions give you the rule and ask what follows in a particular situation.
Either way, check scope in both directions: a principle that's narrower than the case won't cover it, and one that's much broader may commit to things the case never showed. Read the conditions in the principle literally, and make sure the facts actually satisfy every one of them.
Deductive inference: what follows?
Now the mode changes. These stimuli are frequently not arguments at all. They're a set of statements you must treat as true. There's no gap to find and nothing to criticize. The discipline is staying inside what's written.
Must Be True / Most Strongly Supported
"Which one of the following is most strongly supported by the statements above?" Prediction works differently here: rather than predicting an answer, note the combinable pieces, such as two conditionals that chain or a quantity that limits another, and then test each choice against them.
During last week's coastal storm, every buoy in the bay that recorded a wave height above four meters also recorded a wind gust above sixty kilometers per hour. Every buoy that recorded a wind gust above sixty kilometers per hour triggered its automatic warning light. Buoy 3's automatic warning light did not trigger at any point during the storm.
Which one of the following is most strongly supported by the statements above?
- Buoy 3 did not record a wave height above four meters during the storm.
- Had Buoy 3 recorded a wave height above four meters, some other buoy's warning light would have triggered.
- Buoy 3 recorded lower wave heights during the storm than any buoy whose warning light triggered.
- Buoy 3 recorded neither a wind gust above sixty kilometers per hour nor any rainfall during the storm.
- Every buoy whose warning light triggered during the storm recorded a wave height above four meters.
Show the answer
A is correct. The rules chain forward: a wave above four meters means a gust above sixty, and a gust above sixty means the light triggered. So a high wave guarantees a triggered light. The one fact you have about Buoy 3 runs the other way, since its light didn't trigger, so read the chain backward: no light means no qualifying gust, and no qualifying gust means no wave above four meters.
E reverses the rules, treating a triggered light as proof of a high wave; a gust alone could have triggered it. C turns that same reversal into a comparison between buoys that the statements never make. D starts correctly, because no qualifying gust does follow, and then bolts on a claim about rainfall that nothing supports. B invents a relationship between one buoy's readings and another buoy's light.
Two habits pay for themselves on this type. First, when you see a conditional, note its contrapositive immediately. A large share of these questions are solved by running a rule backward, as this one is. Second, be ruthless about choices like D, which pair something that follows with something that doesn't. A choice is only correct if all of it is supported.
Must Be False
The inverse and much rarer: which choice is incompatible with the statements. Same skill, reversed target.
Main Point / Main Conclusion
"Which one of the following most accurately expresses the main conclusion of the argument?" Apply the "why" test in the stimulus, decide on the conclusion, then find the choice that paraphrases it. The reliable trap is the intermediate conclusion: a claim that really is supported by the premises, but exists to support something further. Ask what the argument is ultimately for.
Conclusion / Fill-in-the-Blank
The stimulus ends mid-thought, often after "therefore" or "which suggests that." You supply the conclusion the evidence actually licenses. Predict before looking, and keep the prediction modest: the right answer is normally the most cautious claim the premises can carry, not the most interesting one.
Point at Issue / Point of Agreement
Two speakers, and you identify what they actually disagree (or agree) about. The right answer must be something both speakers address and on which they take opposite positions. Most wrong answers fail one half of that: a claim only one speaker mentioned, or one they'd both endorse.
One argument, four different questions
Here's the thing the type list obscures. The types are not nineteen separate skills. Most of them are different demands made on the same underlying analysis.
Take a simple argument: a study found that houses with professionally landscaped front yards sold for about 10 percent more than houses without, so adding landscaping will increase a house's sale price. The gap is the same no matter what gets asked. The evidence shows only that landscaping and higher prices went together, while the conclusion claims one produces the other, which requires assuming the two groups of houses didn't differ in some other relevant way.
Now watch how the stem changes the job and nothing else:
- Flaw: name the gap. The argument treats evidence of a correlation as though it established causation.
- Weaken: make the gap worse. Those houses were also in more desirable neighborhoods.
- Strengthen: close it. The two groups of houses were comparable in location, size, and condition.
- Necessary Assumption: state what it silently needs. Nothing other than the landscaping accounts for the price difference.
- Evaluate: ask what would settle it. Whether the landscaped houses differed from the others in some further way that affects price.
One piece of analysis, five questions answered. This is why drilling by type in isolation has limits, and why the students who improve fastest are usually the ones who got good at finding the gap rather than the ones who memorized nineteen definitions. (That full walkthrough, on this exact question, is here.)
Recognizing the stem in five seconds
Read the stem before the stimulus. It costs three seconds and tells you what to look for while you read. The tells:
- "if true" — you're in argument evaluation. The choice is new information from outside the stimulus. Strengthen, Weaken, or Paradox.
- "required by" vs "if assumed, allows" — necessary versus sufficient. The single highest-value distinction in the entire section.
- "most accurately describes" — structural analysis. You're describing the argument, not fixing or attacking it.
- "most strongly supported by" — inference; the choices follow from the stimulus. Careful: "most strongly supports" (the other direction) is a Strengthen question. One word, completely different task.
- "the argument proceeds by" or "plays which role" — method or argument part.
That last pair is worth reading twice. Supported by means the stimulus holds up the answer; supports means the answer holds up the stimulus. A surprising number of missed questions are just that reversal.
What to do with this
Don't try to memorize nineteen definitions. Learn the four families, get fluent at spotting which one a stem belongs to, and put your practice time into finding the gap, the skill that all three argument-based families share.
Then track your accuracy by type, because the types are not equally hard for any two people. Some students lose most of their points to Parallel Reasoning and Method questions; others are fine there and bleed on Necessary Assumption. Those are different problems, and studying "Logical Reasoning" as one undifferentiated block is how people spend three months practicing the types they were already good at.
Practice this with Forge180
Forge180 tracks every one of these question types individually, so you can see which ones are actually costing you points instead of guessing.
- An AI tutor that coaches the analyze, predict, eliminate process instead of just showing you the answer
- Adaptive drills that pick each question from your actual strengths and weaknesses
- Analytics that show which question types and which skills are moving
Forge180 launches in August 2026. Email [email protected] to hear when it opens.