Parallel Reasoning: Abstracting Structure, Not Topic
Parallel Reasoning has one job: strip the stimulus to its skeleton and match the structure. Topic is a decoy — the answer rarely resembles it.
Parallel Reasoning asks whether an argument has the same logical structure as the stimulus. That is the entire task, and it is the only LR question type where reading the answer choices before understanding the stimulus reliably produces a wrong answer — because the wrong choices are designed to sound like the stimulus while differing in structure.
The stimulus's subject matter is not merely irrelevant, it is actively misleading. A stimulus about physicians and a correct answer about accountants can be a perfect structural match, while a choice that reuses the stimulus's exact vocabulary is usually the trap.
The two kinds of parallel questions
A regular Parallel question asks which argument reasons the same way. A Parallel Flaw question asks which argument contains the same reasoning error — which is harder, because you must first find the error in the stimulus, then find it again in a choice. The error families in our Flaw questions guide apply directly.
The abstraction procedure
You are not summarizing the stimulus. You are reducing it to a form that strips nouns and keeps only the logical operators.
- Strip the content
Replace every noun phrase with a letter. A stimulus about doctors prescribing medication becomes: "All A are B; some A are not C; therefore some B are not C." The letters carry no meaning, which is exactly the point — they prevent you from being attracted by similar subject matter.
- Mark the conclusion and the premises
Write "P1," "P2," "C" in the margin. Parallel questions have more than one premise frequently, and the error is usually in how a conclusion relates to the premises, not in any single one.
- Capture the operator
What connects the premises to the conclusion? All → some → some is a different operator than all → some → all. Quantifier drift is the most common structural difference between choices.
- Handle conditionals separately
If the stimulus contains an if-then structure, translate it explicitly — including the direction. A choice with the correct form but reversed direction is the most common trap on conditional parallels, so check it before anything else. Our conditional reasoning guide covers the translation.
Do not skip the abstraction even for easy stimuli
The abstraction takes about fifteen seconds and prevents the single most expensive error on this type, which is selecting a choice because it resembles the stimulus. Students who skip it report feeling they should have known better — the resemblance felt like evidence, and it is not.
What actually differs between choices
Once you have the skeleton, the differences between answer choices fall into a small number of categories. These are the only things worth comparing.
| Difference | Example | Why it is wrong |
|---|---|---|
| Quantifier | Some vs. all | A stronger claim than the structure supports |
| Conditional direction | P → Q vs. Q → P | Reverses the logic while looking identical |
| Negation placement | Not A vs. A is not B | Changes which term is excluded |
| Conclusion type | Necessary vs. possible | Asserts necessity the stimulus only suggests |
| Number of premises | Two vs. three | Structural mismatch even if each part matches |
| Number of terms | Three vs. four | Subtle but decisive on harder questions |
Topic, vocabulary, and plausibility are not on this list because they never determine a parallel answer.
The quantifier check
All → some → some is valid: if all A are B, and some A are not C, then some B are not C. All → some → all is not — the stimulus does not establish that every B escapes C. Many wrong answers make exactly this move, keeping the same nouns and upgrading a partial conclusion to a universal one.
Parallel Flaw questions
For a Parallel Flaw question the work is doubled: identify the error in the stimulus, then match it. The efficient path is to do the full Flaw procedure first — conclusion, evidence, gap, one-sentence error name — and only then read the choices looking for that error in a different subject.
A useful shortcut: Parallel Flaw choices usually contain the *same* error but in a form that looks slightly different, plus one choice that repeats the error from a different family. If you named the stimulus error in a phrase, you can often pick the match from the phrasing alone.
If the stimulus error is reversal, check every choice's direction
Reversal is the most common Parallel Flaw error, and the choices are built to differ by a single word that flips the logic. Read each choice twice: once for meaning, once for direction.
Pacing
Parallel questions consume more time than any other LR type because candidates read every choice carefully — which is exactly the behavior the type punishes. Set a limit: abstract in fifteen seconds, read the choices once looking for structure, and eliminate on operator mismatch before re-reading anything.
If you find yourself reading a choice twice, the abstraction was probably not detailed enough. Rebuild it rather than re-reading — a correctly abstracted skeleton makes most choices obviously wrong on first pass.
Parallel Reasoning appears a few times per section, typically four to six, and it is one of the types most improved by a single habit change. For how it fits alongside other types, see our guide to the 14 question types by frequency; for timing management across the section, our 3-month study plan covers where to practice this.
Questions we get asked
How many Parallel Reasoning questions are on the LSAT?
Typically four to six per scored LR section, which makes them a meaningful share of the section despite being individually slow. They are also one of the highest-yield types to improve, because the abstraction procedure is mechanical and the mistakes are identifiable.
Should the correct answer be on the same topic as the stimulus?
No, and often the opposite. Topic similarity is the trap the type is built around. A structurally perfect match may be about a completely different subject, and a choice reusing the stimulus's exact subject matter is frequently wrong.
How do I find the right abstraction level?
Remove content but keep every logical operator: quantifiers, conditionals, negations, and the number of premises. If your abstraction dropped a term or a premise, the choices become indistinguishable — that is the signal you abstracted too far.
What makes a Parallel Flaw question harder than a regular one?
You must perform two procedures in sequence: find the logical error in the stimulus, then find it again in a different argument. Candidates who skip the first step and match on superficial resemblance usually pick a choice with the same topic and a different error.
How long should a Parallel question take?
Under two minutes, and much less once the abstraction is habitual. The usual cause of a slow Parallel question is re-reading choices rather than rebuilding the skeleton — if a choice seems ambiguous, the abstraction was incomplete.
How much does the LSAT cost, and can I take it more than once?
Registration is $253 for the 2026–2027 testing year and includes LSAT Argumentative Writing. LSAC limits sittings to five within the current reportable score period and seven over a lifetime, with no per-year cap; cancelled scores count toward both limits. Need-based fee waivers cover registration for eligible applicants.
LSAT® is a registered trademark of the Law School Admission Council, which does not review or endorse this site. This article is educational and is not legal or admissions advice. Format and policy figures reflect LSAC publications available as of October 2026 — confirm against lsat.lsac.org before you register.
Sources
- LSAC — Official LSAT PrepTests and sample questions, LawHub library
- LSAC — LSAT Logical Reasoning overview, lsat.lsac.org
Figures reflect the most recent published data available as of October 11, 2026. Always confirm current policy on lsat.lsac.org.