Reading Comprehension is the largest single sub-block inside the GRE Verbal Reasoning measure, and it is also the section where most candidates between 155 and 162 quietly leak points. The reading itself is rarely the problem. The trouble shows up at the question stem, when a candidate has to choose between an answer that is technically true and one that is actually supported by the passage. Inference questions live precisely in that gap, and they account for a disproportionate share of wrong answers even among test-takers who finish each passage with comfortable time to spare. This piece focuses on the specific tactical layer where GRE Reading Comprehension rewards or punishes a reader, with a method that pulls inference accuracy above 80% for most disciplined candidates.
The 4 RC question families, restated around what they actually test
ETS groups Reading Comprehension items into question types, but the official taxonomy collapses several skills that behave very differently on test day. For preparation planning it is more useful to think in four families that share a common underlying demand.
- Detail-locator questions — these ask what the passage explicitly says. The wrong answers are paraphrases that invert a polarity, swap a cause and effect, or shift a quantifier ("all" becomes "most"). If you can re-read the cited line and point to it, you win.
- Inference questions — the passage does not state the answer. A defensible conclusion must be drawn from two or more textual cues. This is the family that costs 165+ scorers the most points.
- Function and structure questions — these ask why the author included a sentence, switched examples, or opened a paragraph with a concession. The skill is reading authorial intent, not vocabulary.
- Application / extension questions — they take a principle from the passage and ask the test-taker to recognise it in a new scenario. They are rare on a single sitting but show up at least once on most administrations.
Most candidates, in my experience, over-train the first family because it is the easiest to grade in self-study. The second family — inference — is where preparation strategy should concentrate, because it is the only one where high reading speed and high accuracy can both fail at once.
What an inference question actually demands on the GRE
Three features separate a true GRE inference question from a tricky detail question. First, the correct answer is never stated in the passage in those exact words. Second, the four options tend to be pairwise similar — two "soft" inferences and two "hard" ones — so the choice usually turns on whether one additional textual step is supported. Third, two of the wrong answers will be true in the real world but unsupported by the passage, which is the trap that catches readers who bring outside knowledge into the question.
A useful diagnostic is the "two-cue test": before selecting an inference answer, the reader should be able to point to two distinct sentences or phrases in the passage that, taken together, make the conclusion more likely than not. One cue is rarely enough. If you can only justify the answer with a single inference step, you are probably looking at a detail question disguised as an inference, or an answer that is too aggressive.
Selective reading: a passage-level method, not a section-level one
The advice to "read selectively" is given so often that it has lost meaning. On a 12-minute block of four passages, selective reading is not about skipping sentences — it is about deciding, paragraph by paragraph, which sentences carry argumentative weight and which are decorative. The method has three mechanical steps.
- Read the first sentence of each paragraph before reading the passage. This gives you a one-line skeleton of the argument in under 40 seconds.
- Read each subsequent paragraph for the claim, the evidence, and any concession language ("however", "yet", "although", "still"). Decorative elaborations can be skimmed.
- Mark, mentally, the location of any sentence that contains a polarity word, a number, or a name. These are the sentences that 70% of wrong-detail answers misquote.
For most candidates the first two steps alone recover 3 to 5 minutes per block, which is enough to read the harder third passage twice without rushing the question set. The third step is what closes the gap on inference questions, because the cues you need are almost always anchored in those marked sentences.
The 90-second question triage that protects the inference score
Once inside the question set, the failure mode is not the absence of skill but the misallocation of seconds. A working triage looks like this.
| Question stem signal | Likely family | Time budget | Default move |
|---|---|---|---|
| "According to the passage / the author states" | Detail locator | 45 seconds | Re-read the cited line, eliminate paraphrase traps |
| "It can be inferred / most likely" | Inference | 75 seconds | Apply the two-cue test before committing |
| "The author mentions X primarily in order to" | Function / structure | 60 seconds | Read two sentences before and two after the cited phrase |
| "Which of the following would most weaken / strengthen" | Application | 90 seconds | Identify the passage's claim, then test the option against it |
Two tactical notes. The time budgets above assume an unbroken block of four passages; on the second Verbal section of the GRE, the section-level difficulty can shift, and the budgets should compress by 10 to 15 seconds. Second, the table is a triage, not a rule. If a "detail" question cannot be located within 45 seconds, it is probably a disguised inference and should be reclassified.
Common pitfalls and how to avoid them
Five failure patterns account for most lost RC points, and each has a specific counter-move.
- The outside-knowledge trap. The candidate imports a fact they know about the topic and selects an answer that is true in the world but absent from the passage. Counter: write "per passage?" in the margin before selecting any inference answer.
- The single-cue inference. Only one sentence in the passage supports the chosen answer. Counter: apply the two-cue test. If you cannot find a second anchor, the answer is too aggressive — pick the next-best option or leave the item.
- Polarity blindness. The passage says "rare" and the answer says "common". Counter: scan the cited region for quantifiers, modals, and negations before locking in.
- Reading the question stem first. The candidate skims the passage looking for the answer to question 1, loses the skeleton, and runs out of time on question 4. Counter: read the passage once, then answer the questions in order, even if a question looks answerable from a quick scan.
- Treating short passages as easy. The single-paragraph and short-two-paragraph passages are dense and often contain two or three inference questions in a four-item set. Counter: budget at least 90 seconds for the question set, not the 60 that the passage length implies.
How RC scoring actually works, and why inference accuracy matters more than raw count
The GRE Verbal Reasoning section is section-level adaptive: the difficulty of the second module is calibrated to performance on the first. Within each module, every question contributes to a single score, so there is no "easy" or "hard" sub-score to chase. A consequence that many candidates miss: a high raw count of correct answers on easy detail questions does not compensate for losses on inference items at the top of the difficulty range. The test's adaptive logic effectively weights the harder items more heavily, because a candidate who reaches them has already cleared a difficulty threshold.
For most candidates reading this, the practical implication is that 80% inference accuracy on a hard passage is worth more to the final scaled score than 100% accuracy on a soft passage. Preparation time should reflect that. If you are scoring 158 on Verbal practice tests, your marginal gains almost certainly live in the inference family, not in another pass of vocabulary.
A four-week preparation strand for the inference family
For candidates who already have a working vocabulary base and stable timing on Quant, the following four-week strand is usually enough to lift RC inference accuracy from the low 70s to the mid 80s.
- Week 1 — diagnostic. Take one timed RC block from the official GRE practice tests. Tag every wrong answer by family. If inference items account for more than 40% of the misses, the rest of the strand is justified.
- Week 2 — cue-spotting drills. Use untimed passages. Cover the answer choices. For every paragraph, write down the two sentences that carry the most argumentative weight. This is the muscle the two-cue test relies on.
- Week 3 — full timed blocks with post-mortems. Two blocks per session, each followed by a 20-minute review in which the only thing that matters is identifying which wrong answer was "plausible but unsupported" versus "directly contradicted".
- Week 4 — section-level simulation. Two full Verbal sections per session, scored. Track inference accuracy separately from overall RC accuracy. Most candidates finish week 4 with inference accuracy two to four points higher than their week-1 baseline.
The strand works because it trains the habit the inference family actually rewards: returning to the passage, two or three times, with a specific question in mind. Reading faster is a secondary gain that follows automatically once cue-spotting is automatic.
What to drop from your preparation plan
Two common habits, popular in online forums, actively damage inference accuracy. The first is memorising a long list of "high-frequency inference words" such as implies, suggests, and most likely. On the current GRE these stems do not behave differently from each other in any reliable way; the difficulty is in the answer choices, not the stem. The second is reading a passage aloud or re-reading entire paragraphs after every question. Re-reading is expensive, and the time it consumes is borrowed from later passages where the same mistake pattern recurs.
I'd personally drop both of those habits before dropping anything else. They feel productive because they generate effort, but they do not move the underlying skill.
Where this leaves a 165+ Verbal candidate
Reading Comprehension is the section of the GRE Verbal measure where the gap between a 158 and a 165 is widest, and it is also the section where the gap is most closable through method. Detail-locator accuracy tends to be near ceiling for anyone who finishes the section on time, which means the discriminating items are inferences, function, and application. A four-week preparation strand that focuses on those three families — combined with disciplined timing and post-block review — is usually enough to convert a 158 Verbal into a 162-165 range without changing vocabulary study habits at all.
TestPrep İstanbul's diagnostic assessment on the GRE Verbal Reasoning module is a natural starting point for candidates who want their inference accuracy, not just their overall score, measured before they commit to a preparation plan.