Rank 1 by frequency | 612 questions in corpus (24.7% of all questions)
Inference is the single most common RC question type — nearly a quarter of all questions on the section. You're asked to draw a conclusion that the passage doesn't explicitly state but strongly supports: something the author's words and position make reasonable, without being written anywhere verbatim. The correct inference is the smallest defensible step beyond the text — not a restatement, not a leap, but the next thing a careful reader would conclude.
RC Inference asks you to read between the lines — connecting stated premises to an unstated but warranted conclusion. The correct answer is something the passage makes probable or necessary but never comes out and says directly. The skill is seeing what the text entails without confusing what it entails with what it merely fails to contradict.
The core cognitive demand is conservative extrapolation. The correct inference extends the passage's logic by the smallest defensible step — recognizing a logical consequence of stated facts, identifying what an author's position implies, synthesizing information from two different paragraphs into a conclusion neither states alone, or noticing what a comparison or contrast implies about the items being compared. A correct inference can be defended by pointing to specific passage text.
A key distinction: the passage must support the answer, not merely permit it. Plenty of things are consistent with what the passage says without being supported by it. The LSAT rewards the reader who holds the line between "warranted" and "speculative."
Inference is the most diverse RC question type in stem phrasing, with six distinct subtypes. Each one signals a slightly different cognitive task — and spotting the variation up front changes what you're hunting for in the passage.
Variation A — "It can be inferred." 142 questions (23%). "It can be inferred from the passage that..." or "Which one of the following can most reasonably be inferred from the passage?" The broadest framing — the inference can come from anywhere in the passage and no specific viewpoint is targeted.
Variation B — "The passage suggests / most strongly suggests." 120 questions (20%). "The passage most strongly suggests that..." "Suggests" signals a softer inference — one that may require reading tone, emphasis, or implication rather than strict logical entailment.
Variation C — "The author would most likely agree." 69 questions (11%). "The author would be most likely to agree with which one of the following?" Requires building a model of the author's worldview and then testing each answer against it. The correct answer is consistent with the author's stated views without being explicitly stated by the author.
Variation D — "Most strongly supported by the passage." 22 questions (4%). "Which one of the following is most strongly supported by the passage?" The word "supported" suggests the passage provides evidence for the correct answer — making the subtype feel closer to Specific Reference — but the answer is still not directly stated.
Variation E — "Author would believe / contend / hold." 12 questions (2%). "The author would be most likely to contend that..." Narrower than "would agree" — targets a specific belief or contention, often about a topic the passage discusses tangentially.
Variation F — "Given the information in the passage." 16 questions (3%). "Given the information in the passage, which one of the following...?" The "given the information" framing signals that the inference should be closely tied to specific stated facts, not general impressions.
Unlike Main Point, Inference is rarely something you can fully predict before looking at the choices — the inference space is too wide. The method instead emphasizes disciplined evaluation: test each choice against what the passage actually supports.
Step 1 — Locate the relevant passage territory. If the stem names a topic, a person, or a view, find the sentences that discuss it. If the stem is open ("it can be inferred..."), you'll evaluate choices against the whole passage. Either way, before you start judging choices, know where in the text the evidence will come from.
Step 2 — Test each choice against the Support Test. For every choice, ask: can I point to specific passage text that warrants this? Not just text that's consistent with it — text that supports it. If you can't find the warrant, the choice is out. If you can find it but it requires a stretch, the choice is probably wrong.
Step 3 — When two look close, pick the more conservative one. The correct inference goes exactly one step beyond the text, not two. If you're torn between a boldly worded choice and a cautiously worded one, the cautious one is almost always correct. The LSAT rewards the inference that's the smallest defensible extension of the passage.
Inference correct answers average 16.3 words — shorter than Main Point answers because they state a single inference rather than a full thesis. The construction has two features you'll see again and again.
First, moderate language. Phrases like most likely, probably, tends to, in some cases, or can dominate correct answers. Absolute language (always, never, only, must) is rare in correct inferences because the passage rarely supports absolutes.
Second, staying in the passage's domain. The correct answer talks about what the passage talks about, at roughly the specificity the passage uses. Inferences that generalize to broader domains ("therefore all such theories...") or that transfer to new contexts the passage never raised are almost always traps.
Often the correct answer is a paraphrase combining ideas from different passage locations — a premise in paragraph 1 combined with a claim in paragraph 3 yields something neither paragraph states alone. This is the inference's natural sweet spot.
Every wrong Inference answer exploits a specific error the test-taker might make. The shapes repeat — once you can name them, elimination is faster.
Trap 1 — Overstatement / Too Extreme. Takes a supported idea and pushes it beyond what the passage warrants. If the passage says "some critics question the theory," a wrong answer says "the theory has been widely discredited." The direction is right; the degree is wrong. Defense: check quantifiers — some is not most, question is not reject.
Trap 2 — Outside Knowledge / Scope Shift. A claim that's true in the real world but that the passage doesn't support, or a claim about a topic the passage didn't raise. Defense: ask whether the passage actually discusses this — not whether it sounds right in general.
Trap 3 — Wrong Attribution. Assigns a view to the wrong entity: the passage describes a critic's view, and the wrong answer attributes that view to the author. Defense: track whose view is whose as you read — especially in passages with multiple positions.
Trap 4 — Not Contradicted, But Not Supported. A claim that's consistent with the passage but that the passage provides no actual evidence for. Test-takers pick it because nothing in the passage rules it out. Defense: remember that the passage must support the answer, not merely permit it.
Inference's base difficulty is 3. It drops to 2 when the inference requires combining only two adjacent sentences, or when the stem directs you to a specific paragraph.
Difficulty rises to 4 when the inference requires synthesizing information from three or more passage locations, or when the stem asks about a view the author doesn't directly express — forcing you to extrapolate the author's position from scattered evaluative signals. It reaches 5 on comparative passages requiring cross-passage synthesis, or when the correct answer uses qualified language that's difficult to distinguish from a distortion. The hardest Inference questions almost always have a correct answer whose hedging phrase (tends to, in some cases, generally) is the thing that separates it from a too-strong close-looking wrong answer.
Inference questions appear on both passage types at roughly the rate the passage type appears: 544 single-passage questions (89%) and 68 comparative-passage questions (11%). On comparative passages, inference questions often ask what both authors would agree on, what one author's view implies about the other's subject, or where the two perspectives diverge. These are among the harder inferences because they require holding both passages' positions in mind simultaneously and finding unstated implications at the intersection.
vs. LR Inference. LR Inference ("Must Be True") requires strict provability — the correct answer has to follow with formal-logic-level certainty from a short, tight stimulus. RC Inference allows reasonable extrapolation from a longer, looser passage. Something that would be too weak for LR can be correct on RC; something that would be too strong for RC can be correct on LR. Calibrate accordingly.
vs. Specific Reference. Specific Reference asks what the passage states — the answer is a paraphrase of text you can point to directly. Inference asks what the passage implies — the answer extends the passage rather than restating it. The "most strongly supported" variation sits on the boundary but still asks for something unstated.
vs. Author's Attitude. Attitude questions ask about the author's evaluative stance (approving, skeptical, neutral). Inference can incorporate attitude but more often asks about factual or analytical conclusions the passage supports.
Inference stems average 21 words — longer than Main Point or Primary Purpose stems because they often specify whose view or which part of the passage to draw from. Recognizing the stem variation immediately tells you whether you're looking for a logical consequence, an author's likely agreement, or something the passage merely suggests.