How the NCLEX CAT Exam Works (and How to Beat It)

What CAT Is and Why It Matters

The NCLEX-RN uses Computer Adaptive Testing (CAT). It isn’t a fixed exam where everyone gets the same 150 questions. It’s an algorithm that builds a unique exam for you in real time, question by question, based on how you’re performing.

This matters because if you don’t understand how the algorithm evaluates you, you’ll spend mental energy during the exam on the wrong things. You’ll panic at question 86 because you’re “still going.” You’ll try to read into whether questions feel harder or easier. You’ll second-guess yourself based on question count rather than question content.

Understanding the CAT algorithm won’t change a single question you’re given, but it can take exam mechanics off your mind. When you don’t have to spend energy decoding the test, you can put all your focus where it belongs: on the clinical content in front of you.

How Computer Adaptive Testing Works

The CAT algorithm follows a specific logic sequence. Here’s what happens from the moment you start:

1. Your first question is relatively easy

The exam doesn’t know anything about you yet. It starts everyone with an item of relatively low difficulty and adapts from there. The algorithm needs a first data point to start building your profile.

2. Answer correctly and questions get harder; answer wrong and they get easier

After each response, the algorithm recalculates. Get one right, and the next question will be slightly more difficult. Get one wrong, and the next will be slightly easier. This isn’t a one-to-one trade, because the algorithm uses all your previous answers, not just the last one, but the general principle holds. It’s trying to zero in on the boundary of what you can and can’t do.

3. A logistic model estimates your ability level in real time

Behind the scenes, the NCLEX uses Item Response Theory (IRT), specifically a logistic model that places your ability on a continuum. Every question in the bank has a pre-calibrated difficulty value. As you answer, the algorithm plots your performance against these known difficulty values to estimate where you fall on the ability scale.

Think of it like a thermometer. Each question is a measurement. Early on, the readings swing widely. But as you answer more questions, the algorithm narrows its estimate. By question 60 or 70, it usually has a fairly stable read on your ability. The remaining questions are about achieving statistical confidence in that estimate.

4. The algorithm needs 95% confidence to stop

This is the key threshold. The exam doesn’t stop when it “thinks” you’ve passed or failed. It stops when it’s 95% statistically confident that your ability level is definitively above or definitively below the passing standard. That’s a high bar, and it’s why the exam length varies so much from person to person.

If your performance is clearly above (or clearly below) the standard, the algorithm reaches 95% confidence quickly, sometimes by question 85. If you’re right on the border, it may take all 150 questions to make a determination. Both scenarios are completely normal.

Question Count: 85 Minimum, 150 Maximum

Let’s put the question count anxiety to rest right now.

• Minimum: 85 questions (including 15 pretest/unscored items)
• Maximum: 150 questions (including 15 pretest/unscored items)
• Most test-takers finish between 85 and 120 questions

Getting 85 questions is neither automatically good nor automatically bad. It means the algorithm reached 95% confidence in its decision early, and that decision could be pass or fail. If you were consistently performing above the passing standard from the start, or consistently below it, the algorithm didn’t need more data. That’s the system working as designed.

Getting 150 questions isn’t automatically good or bad either. It means the algorithm struggled to reach confidence. You were hovering near the passing standard, and it needed every available question to make a call. You can absolutely pass at 150 questions, and you can absolutely fail at 150 questions.

The number of questions tells you one thing: how quickly the algorithm made up its mind. It tells you nothing about the outcome. Question count does not predict pass or fail.

Time Limit: 5 Hours Total

You have 5 hours from the moment you sit down. That includes:

• The pre-exam tutorial (skip it if you’ve practiced — you already know the interface)
• All 85-150 questions
• Two optional scheduled breaks: one after about 2 hours and one after about 3.5 hours of testing
• Any unscheduled breaks you take (these count against your exam time)

The math works out to a little under 2 minutes per question if you get all 150, and more if your exam ends sooner. Some questions take 30 seconds. Some, especially NGN item types like matrix or bowtie questions, take 2-3 minutes. For a per-question pacing target, see Exam-Day Strategies below.

Running out of time is rare, but it does happen. If you hit 5 hours before completing the minimum 85 questions, you fail automatically because the algorithm didn’t have enough data. If you’ve passed the minimum and time runs out, the algorithm bases its decision on all the questions you answered.

The Passing Standard

The NCLEX passing standard is not a percentage. You don’t need to get 70% right, or 80% right, or any fixed percentage. The passing standard is a point on the ability scale: a difficulty threshold that represents minimum competence for safe entry-level nursing practice.

How the passing standard is set

Every three years, the NCSBN conducts a practice analysis. They survey thousands of newly licensed nurses and their supervisors to determine what entry-level nurses actually need to know and do. Based on that analysis, they calibrate the passing standard to reflect current practice demands.

In December 2022, NCSBN’s Board of Directors voted to keep the NCLEX-RN passing standard unchanged at 0.00 logits, effective through March 31, 2029.

What “above the passing standard” means in practice

To pass, you need to demonstrate that your ability level, as estimated by the algorithm, is consistently above this threshold. That means you need to be answering questions at and above the passing difficulty level correctly more often than not.

You don’t need perfection. You need consistency. The algorithm expects you to miss some hard questions. It expects you to get some easy ones right. What it’s looking for is a pattern: when you’re given questions at or above the passing standard, can you handle them?

How Difficulty Levels Work

As noted above, every scored question on the NCLEX has a pre-calibrated difficulty value. That value was determined through pretesting: each new question first appears as one of the 15 unscored pretest items on live exams, and candidates’ performance places it on the difficulty scale before it ever counts toward a score.

What the difficulty levels mean to the algorithm

“Above passing level” questions: These are calibrated above the passing standard. The algorithm serves them to test whether you can perform at a higher level when it estimates you’re competent. Getting a reasonable number of them right is a positive signal to the algorithm.

“Below passing level” questions: These are calibrated below the standard. The algorithm uses them to check whether you can at least meet the minimum. Everyone gets some below-standard questions, especially early in the exam, so they aren’t cause for panic on their own. Getting them wrong, though, lowers the algorithm’s estimate of your ability.

“Near passing level” questions: These cluster right around the passing standard. They’re the tiebreakers: if you’re on the border, these questions carry significant weight in the final determination.

You can’t reliably tell these categories apart while you’re testing (see “Don’t change your approach based on perceived difficulty” under Exam-Day Strategies). Treat this as a description of how the algorithm works, not something to monitor during the exam.

Every question counts, so don’t fade at the end

The algorithm uses all your answers. Every scored response counts toward the final ability estimate, and recent questions get no extra weight: there is no ‘last 20’ or ‘last 60’ rule. But a tired, careless stretch at the end still counts against you like any other mistakes.

So don’t fade at the end. If you’re at question 100 and feeling exhausted, take a break. Splash water on your face. Eat your snack. Come back sharp. Those final questions aren’t the time to start guessing because you want it to be over.

What Happens at the End

Your exam stops for one of three reasons:

1. The 95% confidence rule is met. The algorithm is statistically confident your ability is either above or below the passing standard. This can happen at any point between question 85 and 150.
2. You reach 150 questions. The algorithm ran out of questions. It now makes a final determination based on your last ability estimate relative to the passing standard — whichever side you’re on, that’s the call.
3. You run out of time. If you’ve completed at least 85 questions, the algorithm bases pass/fail on all your scored responses. If you haven’t reached 85, you fail.

The 95% confidence interval — explained simply

Imagine a bell curve centered on the algorithm’s best estimate of your ability. The 95% confidence interval is the range that captures where your “true” ability almost certainly falls. If the entire bottom edge of that range is above the passing standard, you pass. If the entire top edge is below the passing standard, you fail.

If the range straddles the passing standard, meaning the algorithm can’t rule out that you might be above or below, it keeps asking questions to narrow that range. At 150 questions, if it’s still straddling, your point estimate (the center of the bell curve) determines the outcome. This is why borderline candidates get the maximum number of questions.

Exam-Day Strategies

Here’s how to use this knowledge on exam day.

Don’t track question count during the exam

It’s tempting, but don’t. You’ll start interpreting what the count “means,” and it will distract you. The question counter is there for time management, not for reading tea leaves about your performance. Glance at it for pacing. Nothing more.

Treat every question like it’s the deciding question

Because in a sense, it is. Every scored response feeds your final ability estimate. There’s no throwaway question. Don’t mentally check out because you think you’ve “probably passed already” or “probably failed already.” Neither assumption is reliable, and both lead to careless answers.

Don’t change your approach based on perceived difficulty

If you feel like questions are getting easier, that doesn’t necessarily mean you’re failing. You might be getting pretest (unscored) items, which can be any difficulty. You might be misjudging difficulty: a question can feel easy and still be calibrated above the passing standard. Stick to your test-taking strategy regardless of how the questions feel.

Use the breaks

You get an optional scheduled break after about 2 hours of testing and another after about 3.5 hours. Take them. Both of them. Stand up. Walk to the restroom. Eat something with protein and complex carbs. Drink water. Your brain has been making high-stakes clinical decisions for over an hour, and it needs a reset.

Students who skip breaks to “get it over with” tend to make more errors in the back half of the exam. Don’t sacrifice accuracy for speed.

Time management: aim for about 1.5 minutes per question

Averaging about 1.5 minutes per question keeps you safely on pace even if your exam runs to 150. Straightforward multiple-choice questions should take 30-45 seconds, which leaves room for slower NGN items (matrix, bowtie, highlight).

If you’re spending more than 2 minutes on a standard question, make your best choice and move on. There’s no going back, so deliberating endlessly doesn’t help. Don’t rush, and don’t dawdle: read the question, eliminate what you can, commit to your answer, and move on.

Read every word of the question stem

CAT questions are precision instruments. Every word is intentional. “Priority” means something different from “appropriate.” “Initial” means something different from “most important.” “The nurse should first” is a different question than “which action is correct.” If you misread the qualifier, you’ll pick a defensible answer that isn’t the best answer, and a standard multiple-choice item is scored simply right or wrong.

Common Myths Debunked

“If the exam shuts off at 85 questions, you definitely passed.”

False. As explained under Question Count above, stopping at 85 means the algorithm was 95% confident in its decision, and that decision could be pass or fail. The shutoff at 85 tells you the algorithm was confident, nothing more.

“The last question determines pass/fail.”

False. The algorithm doesn’t look at a single question. It uses your entire response pattern to generate an ability estimate with a confidence interval. The last question is one data point among 85 to 150. Getting the last question wrong does not mean you failed. Getting it right does not mean you passed.

“If questions are getting harder, you’re passing.”

Oversimplified to the point of being misleading. Harder questions can be a good sign, since the algorithm may be testing above the passing standard because it estimates you’re competent. But you can’t reliably judge question difficulty while taking the exam. A question about a topic you know well feels easy even if it’s calibrated as hard. A question about a topic you’re weak in feels hard even if it’s calibrated as easy. Perceived difficulty is not calibrated difficulty. Stop trying to read the algorithm and focus on the clinical content.

“You need to get a certain percentage right to pass.”

False. There is no fixed passing percentage. The NCLEX is criterion-referenced, not norm-referenced. You’re measured against the passing standard, not against other test-takers and not against a percentage threshold. A candidate who gets 60% of very hard questions right may pass, while a candidate who gets 80% of easy questions right may fail. It’s about the difficulty level of questions you can handle, not raw accuracy.

“SATA questions are always harder.”

False. Select All That Apply questions have difficulty values just like any other question type. Some SATA questions are calibrated below the passing standard. Getting a SATA question doesn’t mean the algorithm thinks you’re passing. With the NGN scoring model, partial credit is now possible on many item types, which is another reason this myth doesn’t hold.

Preparing for the CAT Format Specifically

Knowing content is necessary. Knowing how to perform under CAT conditions is what separates prepared students from overwhelmed ones.

Practice under adaptive conditions

Fixed-length practice tests don’t replicate the CAT experience. When every practice quiz is 50 questions, you never train for the psychological uncertainty of not knowing when the exam will end. Use an adaptive practice tool that varies question count and difficulty. Get comfortable with ambiguity, because on test day you won’t know if the next question is your last.

Build stamina for the long exam

If your exam goes to 130 or 150 questions, you’ll be sitting for 3-4 hours of active testing. That’s an athletic event for your brain. In the weeks before your exam, do full-length practice sessions. Don’t break a 100-question set into four sessions of 25. Sit down. Do all 100. Train your focus to hold.

Practice with all NGN question types

The current NCLEX includes Next Generation NCLEX item types — extended multiple response, extended drag-and-drop, cloze (drop-down), matrix/grid, highlight, and bowtie, plus trend-style data questions (trend is a presentation style rather than a separate NCSBN item type). If you’ve only practiced standard multiple choice and SATA, you’re underprepared. These item types require different cognitive strategies and often take longer to answer. Familiarize yourself before exam day so the format doesn’t cost you time or composure. See our complete guide to NGN question types for a full breakdown.

Study rationales, not just answers

The CAT algorithm doesn’t care if you memorized a fact. It cares whether you can apply clinical reasoning to novel scenarios. The best way to build that reasoning skill is to study rationales, especially for questions you got wrong. Understand why the correct answer is correct. Understand why each distractor is wrong. That reasoning process is exactly what the NCLEX is testing.

Practice Questions

Here are two questions representative of what you’ll see on the NCLEX. Notice the difference in difficulty — the CAT algorithm serves you questions across a range to triangulate your ability level.

The Bottom Line

The CAT algorithm isn’t your enemy. It’s a measurement tool doing exactly one thing: determining with statistical confidence whether you meet the minimum standard for safe entry-level nursing practice. You don’t beat it by trying to game it. You beat it by showing up prepared, with strong clinical knowledge, practiced test-taking skills, and the stamina to perform consistently from question 1 to question 85 or 150.

So stop worrying about question count, stop trying to gauge difficulty in real time, and stop reading forums about what the last question “means.” Know the content, practice under realistic conditions, and focus on what’s in front of you, one question at a time.

For a structured study approach, start with our NCLEX-RN Study Guide and pair it with daily practice from the question bank.