“Most”: The Most Important Word on the LSAT

By Jeremy LaRusso · Updated October 2026

If I could get students to notice one word on the LSAT, it would be most. It appears throughout Logical Reasoning and Reading Comprehension: in the passages, the questions and the answer choices. It sounds casual in everyday English, so students tend to read right past it. On the LSAT, it changes what a sentence claims, and in my experience it decides more questions than any other single word.

The word is most important on Inference questions, which ask what you can conclude from the stimulus. The stimulus is the short passage that a Logical Reasoning question is about. One of the most common wrong answers on the test is an answer that says most when the stimulus never stated a majority. The word is also extremely relevant on Strengthen and Weaken questions. Strong language is wording that claims a lot, such as all or most. Weak language is wording that claims little, such as some or may. A correct answer on a Strengthen or Weaken question often needs strong language, and most is strong language.

This article explains what the word means and how it differs from the most. Then it explains how the word is tested on each question type: Inference, Strengthen and Weaken, Assumption, Parallel Reasoning, Flaw and Reading Comprehension. Every example was written for this article. None is an LSAT question.

What “most” means

On the LSAT, most means more than half: more than 50 percent of the group.

It does not mean “almost all” or “the vast majority.” Fifty-one out of a hundred is most.

It also does not mean “most but not all.” In conversation, “most of the guests left early” suggests that some guests stayed. On the LSAT, it does not suggest that. If every guest left early, it is still true that most of the guests left early.

Real LSAT questions test this. Suppose a stimulus says that all of a club’s members voted. Then an answer saying that most of the club’s members voted is supported. I have watched students eliminate that answer because the stimulus said all and the answer said most. But most does not support all. If you know that most members of a group have some feature, you do not know whether all of them have it.

Some other expressions mean the same thing as most, and some do not.

You need one more fact about the meaning of most, because Necessary Assumption questions test it. On those questions you check an answer by supposing that it is false. If “most of the guests left early” is false, that does not mean that none of the guests left early. It means that half or fewer of the guests left early. With 100 guests, 50 or fewer left early.

“Most” and “the most”

The LSAT also uses the word with a second meaning, and that meaning is not “more than half.” The most means more than each of the other options has.

The most votes means more votes than any other candidate. Most of the votes means more than half of all the votes. A candidate can have the most votes without having most of the votes. An argument is flawed if it concludes that a candidate has most of the votes because she has the most votes. The LSAT tests this flaw, and the Flaw section explains it.

“Most likely” can mean likelier than any other possibility

When most likely compares several possible outcomes, it means likelier than each of the others. The chance of the most likely outcome can be far less than 50 percent. In a race with nine horses, the horse with the best chance of winning might have a 25 percent chance. That horse is the most likely winner, and it will probably lose.

Most likely does not always compare several possible outcomes. “The roads will most likely be icy tomorrow morning” does not mention any other outcomes. In that sentence, most likely means probably. Read the whole sentence to determine which meaning it has. Probably and more likely than not always mean a chance greater than 50 percent.

The two modes, briefly

How you evaluate most in an answer choice depends on what kind of question you have. I sort LSAT questions into two kinds, which I call modes. The language strength article explains the modes completely. Here is a short explanation.

• Mode 1: you work only with the information you were given. The answer does not add a new fact. Inference, Necessary Assumption, Flaw, Main Conclusion and Parallel Reasoning questions are Mode 1. So is nearly every Reading Comprehension question.
• Mode 2: the answer choice is a new fact that is not in the stimulus. You add it to the argument and accept it as true. Strengthen, Weaken, Sufficient Assumption and Paradox questions are Mode 2. (Paradox questions ask you to explain two facts that seem to conflict.) So are principle questions that ask which principle most helps to justify the argument, and principle questions that ask which answer justifies applying a principle. A Most Logically Completes question ends with a blank that the answer fills. It is Mode 2 when the word or phrase before the blank introduces a reason, such as since, because, for, as or on the grounds that.

Here is the test. Are you adding the answer choice to the argument as a new fact? If you are, the question is Mode 2. If you are not, the question is Mode 1.

The rule about most is not the same on every question type. On Inference questions, an answer can say most only if that quantity was stated to you. On Mode 2 questions, the quantity does not have to be stated to you. An answer that says most adds strong language to the argument, and the correct answer often needs strong language. The other Mode 1 question types have their own rules. On a Necessary Assumption question, ask whether the argument requires the answer. On a Flaw question, ask whether the answer accurately describes what is wrong with the argument. On a Parallel Reasoning question, ask whether the answer matches the argument. On a Main Conclusion question, ask whether the answer states the author’s conclusion with language exactly as strong as the author’s. This article explains each of these.

“Most” in Inference questions

Inference questions give you information and ask what you can conclude from it. Must Be True and Most Strongly Supported questions in Logical Reasoning are Inference questions. So are most Reading Comprehension questions. On all of them, the correct answer cannot claim more than the stimulus or passage states.

The unstated quantity appears on Must Be True and Most Strongly Supported questions and on Main Conclusion questions. In Reading Comprehension, it appears on Main Point questions, on questions that ask what the passage suggests or implies, and on questions about someone’s opinion. The opinion can be the author’s. It can also be the opinion of anyone else the passage describes, such as a historian or a critic.

When the stimulus states a quantity, every weaker quantity is also true. If all members of a group have a feature (and the group has at least one member), then most of them have it. If most of them have it, then some of them have it. So the correct answer can be as strong as the stated quantity, or weaker. It can never be stronger. I tell students to think of a less-than-or-equal-to sign: the strength of the answer has to be less than or equal to the strength of what the stimulus stated. If the stimulus states most, the correct answer might say most, or it might say some. If the stimulus states only some, many or often, the correct answer will not say most, because none of those words states a majority.

Read the whole stimulus before you eliminate an answer. A majority can be stated without the word most. If the stimulus says that 70 percent of the members voted, then most of the members voted.

You answer Most Strongly Supported questions almost exactly the way you answer Must Be True questions. The word most is in the question stem, which is the sentence that asks the question. It does not mean you can pick an answer that is merely possible or reasonable. The correct answer is essentially guaranteed by the stimulus. If the correct answer needs an assumption that the stimulus does not state, that assumption is minor and is common sense.

A most statement about a group can also support a probably statement about one member of that group. Probably is a likelihood, not a quantity, so this is a different kind of inference. Suppose the stimulus says two things: most of a hospital’s night nurses work twelve-hour shifts, and Dana is one of its night nurses. Then an answer saying that Dana probably works twelve-hour shifts is supported, if nothing in the stimulus says that Dana is different from the other night nurses. The stimulus makes it likely that Dana works twelve-hour shifts. It does not make it certain. So an answer saying that Dana works twelve-hour shifts, with no probably, claims more than the stimulus supports. And you can conclude probably only because most means more than half. Often would not support probably, because something can happen often without happening more than half of the time.

Main Conclusion: a weaker answer is also wrong

A Main Conclusion question is not an Inference question. It does not ask what you can conclude from the stimulus. It asks which answer states the author’s conclusion. So the correct answer makes the same claim as the author’s conclusion, with language exactly as strong as the author’s. An answer that claims a majority the author did not claim is wrong. An answer that is weaker than the conclusion is wrong too. If the author concluded that most of a club’s members commute by bike, an answer saying that some members do is not the claim the author made, so it is wrong.

Combining “most” statements

Inference questions also test what you can conclude when a stimulus has a most statement and another quantity statement. A quantity statement is a statement with a quantity word, such as all, most or some. The fastest way to find out is to use example numbers. Pick one group and suppose it has 100 members. Read the statements one at a time and work out how many of the 100 each statement describes. For example, if most of the 100 members ski, at least 51 ski. Then check what has to be true.

Follow two rules when you use numbers.

• One counterexample disproves an inference. One example in which the inference is true does not prove it. A counterexample is one set of numbers that makes everything you were given true and makes the inference false. If you find one, the inference does not have to be true. To show that an inference is true in every case, you need the reason it is true. The table, or the example after it, gives the reason for each inference.
• Do not assume that two groups are the same size. Suppose a stimulus is about a club’s runners and its swimmers. If you suppose there are 100 runners and 100 swimmers, you have assumed the two groups are the same size, and the stimulus did not say that. Sometimes you need numbers for more than one group. A counterexample needs them. That is allowed if every statement in the stimulus is true with your numbers.

In the table, A, B and C each stand for a group.

Compare “most A are B, and most A are C” with “most A are B, and most B are C.” In the first pair, both most statements are about group A, so at least one member of A is in both majorities. In the second pair, one most statement is about group A and the other is about group B, and nothing is guaranteed about A and C.

The row “most A are B, and only a minority of B are A” needs a longer explanation. From those two statements you can conclude that there are more B than A. Neither statement says so directly.

That conclusion does not violate the rule about unstated quantities. Both quantities were stated to you, and together they guarantee the comparison. So an answer that states the comparison is correct. An answer that compares the sizes of two groups is wrong when nothing stated in the stimulus guarantees that comparison.

“Most” in Strengthen and Weaken questions

Strengthen and Weaken questions are Mode 2. You add the answer to the argument as a new fact and accept it as true. So you are not asking whether the stimulus stated most. You are asking whether the answer is strong enough to strengthen the argument or weaken it.

On these questions, an answer that says most is often the correct one. Most of the conclusions you are asked to strengthen or weaken are not claims about every case. They are general claims: that one thing causes another, that something is probably true, or that something is true of most of a group. A most statement describes more than half of a group, so it can make a claim like that clearly more likely or clearly less likely. Some and many do not state what fraction of the group they describe. That is often too little information to make the conclusion clearly more likely or clearly less likely.

Most is not the only way to support probably. Evidence about Ines herself, such as how well she performed in the earlier rounds, could support it too. But suppose the only evidence is the fraction of finalists who are offered positions. Then that fraction has to be more than half to support probably.

On a Strengthen or Weaken question, you are not choosing between a good answer and a slightly worse one. One answer is correct and four are incorrect. An answer is incorrect if its language is too weak to make the conclusion clearly more likely or clearly less likely. It is incorrect even if it has some small effect on the argument.

Strong language is not always required. Sometimes an answer with weak language is enough to weaken a conclusion. Suppose a conclusion says all, none or never. One case that contradicts the conclusion shows that the conclusion is false, so an answer that says some can be correct. Read the conclusion first. Then ask how strong a new fact has to be to make that conclusion clearly more likely or clearly less likely. Paradox questions are similar: an answer with weak language can be enough to explain the two facts that seem to conflict.

“Most” in Assumption questions

There are two assumption question types. Necessary Assumption is Mode 1 and Sufficient Assumption is Mode 2, and you evaluate most differently on each.

Necessary Assumption

A Necessary Assumption question asks for something the argument requires. The answer is not a new fact that you accept as true, so the question is Mode 1. But it is not an Inference question. The correct answer does not have to be stated in the stimulus. So a correct answer can say most even when nothing in the stimulus states a majority, if the argument requires a majority.

That check is called the Negate Test. To negate an answer is to suppose that it is false. If the argument no longer supports its conclusion, the argument requires the answer. That is what necessary means. Negate most correctly: if a most statement is false, then half or fewer is true. None does not have to be true.

Sufficient Assumption

A Sufficient Assumption question is Mode 2. You add the answer to the argument as a new fact, and the answer has to guarantee the conclusion. A most statement by itself cannot guarantee that every member of a group has a feature. It also cannot guarantee that one particular member has the feature. But a most statement together with a premise can guarantee a conclusion.

Read the festival argument again as a Sufficient Assumption question. Now Answer 2 is correct and Answer 1 is wrong. Answer 2 says that most of this year’s volunteers live near the festival grounds. The argument says that most of this year’s volunteers enjoyed the work. Two majorities of the same group add up to more than the whole group, so at least one volunteer is in both. (The section Combining “most” statements explains this rule.) That volunteer enjoyed the work and lives nearby. The second premise says that this volunteer comes back. So Answer 2 guarantees the conclusion. Answer 1 does not guarantee it, because the volunteers who live nearby might not have enjoyed the work.

Each answer is correct on one of the two questions and wrong on the other. A Necessary Assumption question asks what the argument requires. A Sufficient Assumption question asks what guarantees the conclusion.

“Most” in Parallel Reasoning

A Parallel Reasoning question asks which answer reasons the same way the stimulus does. The quantities have to match. Suppose a premise in the stimulus says most. Then the matching premise in the correct answer says most or a word that means the same. An answer that says some or many there does not match. If the conclusion in the stimulus says probably, the conclusion in the correct answer says probably or a synonym. Checking the quantity and likelihood words is the most reliable way to eliminate answers on these questions. You can often eliminate three answers this way before you compare the reasoning.

Check for synonyms before you eliminate. More than half, a majority, usually and generally all mean the same as most. Many and often by themselves never mean most. Two whole sentences can also mean the same thing. For example, “all of the club’s members ski” and “no member of the club is a non-skier” make the same claim in different words. When an answer uses a different quantity word, first check whether its sentence makes the same kind of claim as the matching sentence in the stimulus. Eliminate the answer only if the two claims differ in quantity.

On a Flawed Parallel question, the stimulus is a flawed argument and the correct answer has the same flaw. Look for this argument, which uses the two statements in the last row of the table in the section Combining “most” statements: “most A are B, and most B are C, so most A are C.” The argument looks correct, but its premises do not guarantee its conclusion. That makes it a likely choice for the flawed argument you have to match.

“Most” in Flaw questions

The word appears less often on Flaw questions than on the question types in the earlier sections, but several common flaws depend on it. When the argument in a Flaw question contains most or the most, check for these flaws first.

• The most treated as most. “More customers chose the Classic plan than any other plan, so most customers chose the Classic plan.” If the company offers five plans, the Classic plan can have more customers than any other plan and still have only 30 percent of the customers.
• Most likely treated as more likely than not. “Of the five finalists for the scholarship, Ortiz is the most likely to win it, so Ortiz will probably win it.” With five finalists, the finalist who is most likely to win can have a 30 percent chance.
• Most treated as certain for one member. “Most of the town’s restaurants close on Mondays, so the Elm Street Grill, one of the town’s restaurants, must be closed this Monday.” The premise supports probably. The conclusion claims certainty.
• Most A are B treated as most B are A. “Most of the region’s award-winning wines are made from local grapes, so most wines made from local grapes win awards.” The premise is about award-winning wines only. There can be far more wines made from local grapes than award-winning wines, so most wines made from local grapes may win nothing.
• Two most statements about different groups combined. “Most of the firm’s partners are litigators, and most litigators work long hours, so most of the firm’s partners work long hours.” The second premise is about all litigators. The firm’s partners who are litigators can all be litigators who do not work long hours.
• Most of the people who answered treated as most of the whole group. “Most of the readers who answered our online poll want a print edition, so most of our readers want one.” The premise is about the readers who answered the poll, and the conclusion is about all of the readers. So the argument assumes that the readers who answered are like the readers who did not answer. Readers who choose to answer an online poll may be different from readers who do not.

In each case, ask what the premise actually supports. Then compare that with what the conclusion claims.

When a Flaw answer contains “most”

A Flaw question is Mode 1. The correct answer describes what is wrong with the argument you were given. Some Flaw answers state a possibility the argument did not consider. Those answers look like correct answers to a Weaken question, so it is easy to forget that a Flaw answer only describes the argument. Compare the two.

Some answers say that an argument overlooks the possibility that something is true. Others say that it ignores the possibility that or fails to consider that something is true. A correct Flaw answer of this kind states a possibility that would weaken the argument if it were true. That gives you a test. Ask: if this possibility were true, would it weaken the argument? If it would not, the answer is wrong. An argument does not have to consider a possibility that would not weaken it.

Use a different test when an answer says that an argument fails to consider whether something is true. That answer states a question, not a possibility, and the question has two possible answers: yes and no. Ask: would either one weaken the argument? Suppose an answer says that the store argument fails to consider whether the customers surveyed are like the store’s other customers. If they are, the argument is not weakened. If they are not, it is weakened. So that answer describes a flaw. If neither yes nor no would weaken the argument, the answer is wrong.

“Most” in Reading Comprehension

The rules of an Inference question apply to most Reading Comprehension questions: the answer has to be stated in the passage or supported by it. So the unstated quantity is as common here as in Logical Reasoning.

Read what the whole answer claims. Suppose an answer says: “The critics believe that most of the letters are forgeries.” That answer is a claim about what the critics believe. It is supported if the passage says that the critics hold that belief. The passage does not have to say that most of the letters really are forgeries. This matters on opinion questions, which can ask about the author’s opinion or about the opinion of someone the passage describes.

Reading Comprehension occasionally has a Strengthen, Weaken or Parallel question. You answer it the way you answer one in Logical Reasoning.

Look especially for the most in answer choices. Passages often praise, criticize and describe things without saying that anything is the best, the worst or the most important. Suppose the passage said “an important part” and an answer says “the most important part.” Or the passage said “a talented painter” and an answer says “one of the most talented painters of the period.” Each answer claims more than the passage did.

The word most is in many Reading Comprehension question stems. There it means the most, as the next section explains. Two kinds of question need more explanation. The first kind asks what the passage most strongly suggests. Those are Inference questions, and the Inference rules in this article apply to them.

The second kind asks what the author values most highly or considers most important. Like any other question, it is answered from what the passage states or supports. Two things usually help you find the answer. One is explicit language: the author calls something the most important, most significant or most valuable. The other is frequency: when the author discusses one thing many times, check whether that thing is the answer. Then read what the author actually says about it. If the author criticizes something many times, that does not show that the author values it highly. If the author praises something many times, that does not show that the author values it more than everything else.

“Most” in the question stem

The word also appears in question stems. A stem can ask which answer most strengthens the argument, which answer is most strongly supported, or which answer most accurately expresses the conclusion. In a stem, most means the most: more than any other answer. It does not mean more than half. But you do not have to rank the answers. One answer is correct and four are incorrect. On a Strengthen or Weaken question, an answer whose language is too weak for that argument is incorrect, even if it has some small effect on the argument. Do not treat it as almost correct. So ask why each answer is correct or incorrect. Do not ask which answer is better.

What the correct answer has to do depends on the stem. On an Inference question, the stimulus has to support the answer, as the Inference section explains. On a Strengthen or Weaken question, the correct answer has to make the conclusion clearly more likely or clearly less likely. It does not have to guarantee that the conclusion is true or guarantee that it is false. On principle questions that ask which principle most helps to justify the reasoning, the correct answer is the principle that supports the conclusion best, and it will either guarantee the conclusion of the argument true or at least strongly suggest that the conclusion is true. If the stem instead asks for a principle that makes the conclusion follow logically, the correct answer has to guarantee the conclusion.

A four-question check

When you see most, ask four questions.

• Which meaning is it? More than half of a group (most of the votes), or more than any other (the most votes)? With most likely, check whether it compares several possible outcomes or means probably.
• What kind of question is it? On an Inference question, was the quantity stated to you? If it was not, you cannot infer it. On a Strengthen, Weaken or Sufficient Assumption question, is the answer strong enough to strengthen the argument, weaken it or guarantee the conclusion? On a Necessary Assumption question, does the argument require it? On a Flaw question, does the answer accurately describe what is wrong with the argument?
• Is there another quantity statement? Check which group each statement is about. If most of the club’s members ski and most of the club’s members hike, at least one member does both. If most of the trees are apple trees and all of the apple trees are grafted, most of the trees are grafted. If most of the partners are litigators and most litigators work long hours, you cannot conclude anything about whether the partners work long hours.
• On a Parallel Reasoning question, do the quantities match? The quantity in each premise of the answer has to match the quantity in the matching premise of the stimulus. The quantity in the answer’s conclusion has to match the quantity in the stimulus’s conclusion.