Multiple-Choice Questions (MCQ)

Overview

The AP CSP multiple-choice section has 70 questions in 120 minutes and counts for 70% of your total exam score, making it the single biggest factor in whether you pass. The AP Computer Science Principles MCQ section includes three question types in this order: 57 single-select questions, 5 single-select questions tied to a reading passage about a computing innovation, and 8 multi-select questions where you choose 2 of the 4 answer options. Every question has four answer choices, and there's no penalty for guessing.

You'll have the Exam Reference Sheet during the exam. It contains both block-based and text-based programming constructs, and the exam's pseudocode might differ from whatever language you learned in class. Get familiar with it before exam day so you know where to find list operations, robot commands, and procedure definitions without burning time.

The other 30% of your score comes from the Create performance task plus written-response questions you answer during a separate 60-minute portion of the end-of-course exam. That's covered in the Create Performance Task guide. This page is all about the multiple choice.

AP CSP MCQ Format: What to Expect

The section is 70 questions, 120 minutes, 70% of your exam score. Here's the breakdown:

Content is weighted by the five big ideas:

Notice that Algorithms and Programming plus Impact of Computing together make up over half the section. You'll spend a lot of time tracing code and reasoning about how computing affects people.

The section also assesses computational thinking practices, with Computing Innovations (28-33%) and Algorithms and Program Development (20-28%) weighted heaviest. Code Analysis sits at 12-19%, Computational Solution Design at 18-25%, and Abstraction in Program Development at 7-12%. Abstraction questions are the rarest but often the trickiest.

Heads up: the AP CSP exam format is stable for now, but College Board has announced an updated course framework arriving in the 2027-28 school year. Everything on this page describes the current exam.

How to Approach the AP CSP MCQ Section

Pace yourself by question type, because the three types demand different amounts of time. Here's a plan that fits the 120 minutes.

Questions 1-30: First 45 minutes

The opening single-select questions tend to test more straightforward concepts (binary conversion, basic vocabulary, simple code segments). Move briskly and bank time. The moment you read a question, categorize it. Is it about code behavior, system design, data representation, or social impact? Each category has its own playbook, and naming the category gets you halfway to the answer.

For code questions, remember two quirks of AP pseudocode that trip people up constantly. List indices start at 1, not 0. REPEAT loops execute exactly the number of times specified. The exam is testing logic, not language syntax.

Questions 31-57: Next 45 minutes

These single-select questions get harder, with longer code traces and multi-part scenarios. One time-saver: if a question asks about one variable, don't trace the entire program. Track only the values that matter for the answer. Under pressure, students waste minutes computing things the question never asks about.

If a question is eating your time, mark it and move on. A marked question you return to beats three questions you never see.

The reading passage: About 15 minutes

The passage set starts around question 58. You'll get a scenario about a computing innovation, often comparing an original system to an upgraded one, usually with flowcharts. Read the five questions first, then the passage. The questions tell you exactly what to look for.

Pay special attention to what changed between the two systems, because that's where most questions live. The official sample passage describes a call center that upgraded from keypad input to voice input and started storing all caller data (phone number, name, address, credit card number) in a database. The questions then probe exactly the things the upgrade changed: what new input data the system needs, what data privacy concerns the database creates, and what benefits or harms the new design introduces.

These questions hit Practices 3 and 5, so expect prompts like "Which of the following is the MOST likely data privacy concern of the upgraded system?" or "Which of the following is LEAST likely to be a benefit?" Watch for LEAST vs. MOST. Misreading that one capitalized word is the most common way to lose these points.

The multi-select questions: Final 15 minutes

The last 8 questions ask you to choose 2 correct answers out of 4 options. Random guessing only gives you a 1-in-6 shot, so work systematically. Evaluate each option independently first. Ask "Is this statement true on its own?" before comparing options. Then check that your two picks make sense together. Correct pairs usually complement each other (two examples of the same concept, two benefits of the same technology, two ways to achieve the same goal). If your two choices contradict each other, one is probably wrong.

The danger here is overthinking. If you've identified two answers that seem correct, they probably are. Don't talk yourself out of them without concrete evidence.

Wrong-Answer Patterns Worth Knowing

AP CSP distractors follow predictable patterns, and recognizing them turns a blind guess into an educated one. These patterns are strategy, not official rules, but they show up over and over in released questions.

For algorithm and code questions, wrong answers represent common logical errors. Expect a distractor showing the off-by-one result, one showing what happens if the loop never runs, and one showing the result of flipped boolean logic. A released question asks you to fix a counting procedure where count ← 0 sits inside the loop, resetting the count every iteration. The fix is moving the initialization before the loop. If you've made that bug yourself in class, you'll spot it instantly.

For Internet and systems questions, wrong answers confuse sibling concepts. Encryption gets swapped with encoding, bandwidth with latency, the Internet with the World Wide Web. A released packet question gives the correct answer as "each packet contains data plus metadata used for routing," with distractors about encryption keys and direct connections. The distractor about establishing a direct connection would describe a traditional phone call, which is exactly the circuit-switching vs. packet-switching distinction the exam loves.

For security questions, know the precise definitions. Multifactor authentication means multiple types of evidence (something you know, have, or are). A released question's wrong answer is "lock the account after failed password attempts," which is a security measure but not a second factor. Phishing means tricking a user into volunteering information (a fake email asking you to "confirm your password"), not exploiting a software vulnerability.

For impact questions, there are right answers grounded in computing principles, not opinions. On a released digital divide question, the correct answer is the school that provides a device to every student, because policies that assume students already own technology widen the divide rather than narrow it.

Concept-Specific Strategies

A few content areas reward a specific mental move. Here's the shortcut for each.

Boolean logic questions often test De Morgan's Laws without naming them. NOT(A AND B) equals (NOT A) OR (NOT B). A released question asks for an expression that's true when someone can drive (age at least 16) but can't vote (under 18). Both (age ≥ 16) AND (NOT(age ≥ 18)) and (age < 18) AND (NOT(age < 16)) work, and recognizing they're equivalent is the whole question.

Binary questions aren't really math questions. They test whether you understand that everything digital is bits. Counting up in binary works like decimal, just carry at 2. The number of bits needed for n unique items is the ceiling of log₂(n).

Binary search questions have a number pattern. A sorted list of 500 elements needs at most about 10 examinations, because each step halves the search space and 2^10 is roughly 1,000. If you see "sorted list" and "maximum examinations," think powers of 2.

Robot grid questions punish assumptions. Check the robot's starting direction (it's often facing right, not up), and remember turns are relative to the robot's current facing, not compass directions.

Parallel processing questions want the minimum total time. Pair up processes to balance the load across processors. With processes taking 60, 30, and 50 seconds on two processors, put 30 and 50 on one processor (80 seconds) and 60 on the other; the answer is 80 seconds, the longer of the two.

Data vs. metadata questions hinge on one distinction. Data is the content itself (pixel colors), metadata describes the content (date, location). Asking "was this photo taken at a particular public event?" is a metadata question because date and location answer it.

Compression questions test a tradeoff. Lossless preserves everything but compresses less; lossy shrinks files more by discarding information humans won't notice is missing.

Creative Development questions reward the human side of software. Correct answers emphasize collaboration, testing, documentation, and iterative improvement.

Common Mistakes

• Treating AP pseudocode like Python or Java. List indices start at 1 in AP pseudocode, and assuming 0-based indexing produces exactly the off-by-one distractor the exam plants. Drill with the Exam Reference Sheet until the notation feels natural.
• Missing LEAST and MOST in the question stem. Reading passage questions frequently ask for the LEAST likely benefit or concern, and answering the opposite question is a free point thrown away. Circle or mentally flag the capitalized word before reading the options.
• Tracing entire programs when one variable matters. If the question asks for the final value of count, ignore everything that doesn't touch count. Full traces burn minutes you'll need for the multi-select questions at the end.
• Comparing multi-select options against each other too early. Two correct answers can look redundant, and students talk themselves out of a right pick because "they can't both be it." Evaluate each option as true or false on its own first.
• Confusing sibling concepts. Privacy vs. security, encryption vs. encoding, phishing vs. malware, Internet vs. World Wide Web. Build precise definitions with flashcards from the AP CSP key terms glossary so distractors built on these mix-ups don't work on you.
• Spending too long on early questions. If you're on question 30 at the 70-minute mark, you're in trouble, since the passage and multi-select questions need real time. Check your pace at the halfway point and skip-and-mark anything that's stalling you.

Practice and Next Steps

The fastest way to improve on the AP CSP MCQ is timed practice with real-style questions followed by honest review of every miss. When you get one wrong, identify the concept being tested, not just the correct letter. Pattern recognition only builds if you're looking for the pattern.

Start with guided MCQ practice to build fluency one question type at a time, then sit for a full-length AP CSP practice exam to test your pacing across all 70 questions. Working through past exam questions gets you used to the exam's exact wording and the Exam Reference Sheet pseudocode. When you finish a practice set, plug your results into the AP CSP score calculator to see how your MCQ performance combines with the Create task toward a final score. For everything else on the exam, the AP CSP exam prep hub has guides for the full test, including the Create performance task that makes up your other 30%.