GMAT Critical Reasoning question types, explained.

Critical Reasoning is the section that punishes students who read it like a passage. The questions are short, the arguments are dense, and almost every wrong answer is wrong for a specific, repeatable reason. There are only eight question types in the entire CR canon. Once you can identify the type from the stem in five seconds, half the battle is won — because each type has a different correct-answer shape, and using the wrong shape on the wrong type is what most students actually do when they think they have a “CR problem.”

This guide walks through all eight, with the recognition phrasing, the trap built into each, and the question I trained myself to ask before looking at the answer choices.

CR isn't reading comprehension. It's a logic test wearing reading-comprehension clothes. The students who score high are the ones who learn to ignore the clothes.

How CR is structured on the GMAT Focus Edition

CR appears inside the Verbal section. Each question gives you a short argument (usually two to four sentences) followed by a question stem and five answer choices. The stem is what tells you which of the eight types you're facing. Train yourself to read the stem first — the argument means different things depending on the type, and reading argument-first wastes half of your processing.

The eight question types

1. Assumption

Recognition phrasing.“Which of the following is an assumption on which the argument depends?” / “The argument relies on which of the following assumptions?”

What it's actually asking. Find the unstated premise the conclusion needs to be true. Not what would support it; what it requires.

The trap.Test-takers eliminate the right answer because it “feels too broad” or “too obvious.” Assumptions usually arebroad and obvious in retrospect — that's precisely why the arguer didn't state them. The narrow, specific-sounding answer is almost always a trap.

The five-second question.“If this answer were false, would the conclusion still hold?” (The negation test.) If negating the answer breaks the argument, it's the assumption.

2. Strengthen

Recognition phrasing.“Which of the following, if true, most strengthens the argument?” / “...most supports the conclusion?”

What it's actually asking.Find the new fact that, if added to the argument's premises, makes the conclusion more likely to be true.

The trap.The trap on Strengthen is answers that restate a premise. A restatement doesn't add evidence; it just repeats what was already there. The right answer always introduces something new.

The five-second question.“Does this answer give me new information that pushes the conclusion uphill?”

3. Weaken

Recognition phrasing.“Which of the following, if true, most weakens the argument?” / “...most calls into question the conclusion?”

What it's actually asking. Find the new fact that, if added, makes the conclusion less likely.

The trap.Test-takers pick answers that attack the premises rather than the conclusion. Premises are stipulated as true in CR; you can't weaken an argument by disputing them. You weaken by showing that even with the premises true, the conclusion doesn't follow.

The five-second question.“Does this answer give a reason the conclusion could be wrong even if the premises are right?”

4. Inference (Must Be True)

Recognition phrasing.“Which of the following must be true based on the statements above?” / “...can be properly inferred?”

What it's actually asking. Find the statement that mustfollow from the premises — a valid deduction.

The trap. Test-takers pick answers that are probably true or likely truebut not necessarily true. CR inference is mathematical: if there's any case where the premises are true and the answer is false, the answer is wrong. “Most likely” is the wrong standard.

The five-second question.“Can I construct a scenario where every premise is true and this answer is false? If yes, eliminate.”

5. Resolve the Paradox

Recognition phrasing.“Which of the following, if true, would best explain the discrepancy?” / “...resolves the apparent paradox?”

What it's actually asking. The argument sets up two facts that seem contradictory. Find the new fact that makes both true at once.

The trap.Answers that explain only one of the two facts. If the answer leaves the contradiction intact, it's wrong, even if it's a true statement about one half of the puzzle.

The five-second question.“With this answer added, are both of the original facts believable simultaneously?”

6. Find the Flaw / Identify the Reasoning Error

Recognition phrasing.“The reasoning in the argument is flawed because...” / “Which of the following best describes the flaw in the argument?”

What it's actually asking. Identify the logical error the arguer made. The conclusion is wrong; you need to name why.

The trap. Answers that describe a differentflaw than the one actually present, or that describe a fault that exists outside the argument (“the survey was small” when the argument never mentions a sample size). The right answer matches the specific reasoning move the argument made.

The five-second question.“Did the arguer commit the move this answer describes?”

7. Strengthen / Weaken the Plan or Argument by Method

Recognition phrasing.“In evaluating the argument, it would be most useful to know whether...” / “Which of the following, if known, would best help evaluate the proposal?”

What it's actually asking. A variant of Strengthen / Weaken. Find the question whose answer(whether yes or no) would meaningfully change your assessment of the argument.

The trap. Test-takers pick questions that sound relevantbut whose answers don't actually change anything about the argument's strength either way.

The five-second question.“If the answer to this question is yes, does my view of the argument shift? If the answer is no, does my view shift in the opposite direction?” Both must be yes.

8. Bold-Face / Role-of-Statement

Recognition phrasing. Two sentences in the argument are in bold. The stem reads something like: “In the argument above, the two boldface portions play which of the following roles?”

What it's actually asking. Identify the structural function of each bold sentence in the argument (premise, conclusion, counterclaim, intermediate conclusion, evidence, etc.).

The trap. The biggest trap is mixing up the argument's conclusion with the arguer's position— especially when the argument acknowledges a counterclaim. Bold-face questions require you to track who is saying what, not just what is being said.

The five-second question.“Does the arguer endorse this sentence or report it as someone else's view?” Apply to each bold sentence independently.

The general approach (works on every type)

Once you can identify the type, the workflow is the same across all eight:

1. Read the stem first.Two seconds. Identify the type and what it's asking for.
2. Read the argument once, slowly. 30-40 seconds. Mentally tag each sentence as premise, conclusion, counterclaim, or context. Underline the conclusion.
3. Predict. Before looking at the answers, say in your own words what the right answer needs to do. This is the single biggest CR habit. People who skip this step get pulled into wrong answers by attractive surface wording.
4. Match each answer against the prediction.Eliminate based on the prediction, not on whether the answer feels true. CR right answers often don'tfeel true on first read — they feel surprising or broader than expected. The prediction is your anchor.
5. If two remain, re-read the conclusion.Almost every CR question between two finalists comes down to which one most directly addresses the conclusion as stated, not the conclusion as you remember it.

The prediction step is what separates 70% CR accuracy from 85%. Most students do every other step and skip this one, then wonder why they keep falling for the persuasive wrong answer.

The five mistakes that cost most CR points

Mistake 1 — Reading the argument before the stem

You don't know what to look for until you know the type. Reading the argument blind doubles the cognitive load and means you usually have to re-read after looking at the stem.

Mistake 2 — Skipping the prediction

Without a prediction, you're comparison-shopping answers. With a prediction, you're grading them. The second is twice as fast and three times as accurate.

Mistake 3 — Treating the answer set as a multiple-choice

Test-takers re-read each answer in turn until one “feels right.” CR rewards a different process: eliminate definitively wrong answers on the first pass, then decide between the two or three that survive. The first pass should take 30 seconds; the decision should take 30 more.

Mistake 4 — Disputing the premises

On Weaken, on Strengthen, on Inference — you cannot argue with the premises. They are stipulated. The whole game is what follows from them. Students who keep wanting to push back on premises lose easy points.

Mistake 5 — Burning time on the wrong type

If you can't identify the type from the stem in five seconds, you don't know the type well enough yet. Drill type-recognition specifically — pull twenty stems out of context and label each one before reading the argument. Once recognition is automatic, your timing improves disproportionately.

How to practice CR

1. One type per day for a week. Pick Assumption. Do twenty Assumption questions across two days. Then Strengthen for two days. Etc. Two weeks of this and your type-recognition is done.
2. Time individual questions from day one.The 1:50 to 2:00 budget is part of the skill. Untimed practice teaches you to over-think.
3. Log every miss with a tag.Was it a type-recognition error, a prediction-skip error, or a comparison error? The same three patterns repeat across hundreds of CR questions. Tag them — you cannot fix what you cannot name.
4. Mix sets after week three. Mixed sets are where pacing falls apart. Make pacing the focus, not accuracy, until your time per question stabilises.
5. Re-do missed questions cold a week later.If you can't reconstruct the right answer without looking, you didn't internalise the lesson. Re-tag and drill that pattern.

The short version

Eight types. Each with a recognition phrase, a five-second question, and a specific trap. Read the stem first, identify the type, predict before looking at answers, match by elimination, decide between finalists by re-reading the conclusion. The students who score high on CR aren't faster readers — they're faster at recognising which game they're playing.

The platform

Zakarian GMAT's CR chapter teaches each type with the recognition phrasing and prediction discipline this guide outlines, and the question bank is tagged by type so you can drill one type at a time. The error log uses the same tagging vocabulary the guide recommends — type recognition, prediction skip, premise dispute, comparison error — so the patterns surface fast. If you want to run the loop without building it from scratch, the diagnostic is free.