AP · United States History · January 29, 2026 · 5 min read
How Many APUSH Practice Questions Should You Do?
By Makon AI Team · Updated July 15, 2026
During ordinary review, aim for 20–40 APUSH stimulus MCQs, two or three SAQs, and one focused DBQ/LEQ rubric task per week. Every two to four weeks, use a longer timed checkpoint. This is a starting range, not an official quota: students with content gaps need more retrieval; students losing rubric points need fewer MCQs and more writing.
The 2026 exam weights 55 MCQs at 40%, three SAQs at 20%, the DBQ at 25% and LEQ at 15%. A plan of “50 multiple choice daily” neglects 60% of the exam. Verify the current APUSH exam format.
Match volume to the error
| Evidence from last set | Next week's practice |
|---|---|
| Facts missing across a period | Daily retrieval timeline + 15–20 targeted MCQs |
| Stimulus misread | 4 small source sets with author/context/claim annotations |
| SAQ examples vague | 3 SAQs, scored part by part |
| DBQ documents summarized | 2 body paragraphs linking documents to claims |
| LEQ lacks evidence | 4 prompt outlines with two specific examples each |
| Timing collapse | One timed section slice after untimed accuracy improves |
A 45-minute practice block
- 12 minutes: answer 8–10 stimulus MCQs.
- 12 minutes: review correct and wrong choices.
- 12 minutes: answer one SAQ or outline an essay paragraph.
- 9 minutes: retrieve the missed content and schedule one transfer question.
This produces fewer counted questions than a speed run, but it trains source reading, content and writing.
Six-week progression
| Weeks | MCQ | SAQ | Essay work |
|---|---|---|---|
| 1–2 | 20/week untimed-to-timed | 2/week | Thesis/context/evidence outlines |
| 3–4 | 30/week in stimulus sets | 3/week | One timed DBQ half or LEQ weekly |
| 5 | One 55-question section | 3 together timed | One full DBQ or LEQ |
| 6 | Targeted weak-period sets | 2 targeted | Opposite essay type + corrections |
Do not use the same released prompt twice as proof of growth; memory changes the task.
Count reviewed losses, not completed questions
Track:
- correct with evidence;
- correct but guessed;
- wrong content;
- wrong source interpretation;
- wrong chronology/reasoning;
- unanswered.
If 30/40 are “correct” but eight were guesses, your secure accuracy is closer to 22/40. Review uncertain correct answers.
Choose a volume profile
Content-rebuild week
Use 15–20 MCQs from one or two periods, two SAQ parts, and one essay outline. Pair each stimulus set with a blank-page timeline or cause/effect map. The smaller question count leaves time to rebuild missing historical knowledge.
Source-analysis week
Use four short stimulus sets from different periods. Before answering, label author/creator, historical situation, audience, purpose, and claim or visual trend. Add two SAQs built around sources. The goal is recognizing how evidence works, not memorizing the topic of each document.
Writing-repair week
Use 15–25 maintenance MCQs, then spend most time on the missing rubric row. Write three thesis/context pairs, two DBQ body paragraphs, or four LEQ evidence-and-reasoning outlines. Finish with one timed essay segment after untimed accuracy improves.
These profiles prevent a universal weekly number from hiding the actual need.
Worked allocation from one practice set
Suppose a student answers 35 of 45 stimulus MCQs correctly, but six correct answers were uncertain. Four of the ten misses use accurate content from the wrong period, and three misread the author's purpose. The secure result is 29 of 45, with chronology and sourcing as the dominant gaps.
The next week could include 20 MCQs in small source sets, daily five-minute period anchors, two source-based SAQ parts, and two DBQ sourcing sentences. Assigning 100 random MCQs would dilute both problems. After repair, use fresh documents from different periods to verify that chronology and sourcing improved.
Review the question budget on Sunday
Ask:
- Did the planned questions receive full review?
- Which period × skill cell produced the most losses?
- Did an essay point improve on a new prompt?
- Did timing reduce accuracy or completion?
- What volume can fit next week without deleting writing or sleep?
Increase volume only when review remains complete and the student needs endurance or broader sampling. Decrease it when explanations are skipped, the same mistake repeats, or school workload makes the plan unsustainable.
What counts as a new question?
A prompt previously read, discussed, or scored is useful for correction but weak as proof of growth. Use unseen official-format items for checkpoints. If secure questions are limited, preserve some stimulus sets and released prompts for later measurement while using textbook or teacher practice for learning.
Respect AP Classroom security and do not seek leaked items. Released College Board materials provide enough public rubric evidence to learn the task honestly.
College Board publishes recent APUSH FRQs, scoring guidelines and samples. Teachers can assign secure AP Classroom questions. Respect material security.
Makon's APUSH mistake review supplies the six-tag system, the progress tracker combines period and skill, and the format guide shows full-section timing.
Makon action: Allocate next week's question budget from the last set: one MCQ range, an exact number of SAQ parts and one named essay rubric row. A question does not count as complete until its uncertainty or error is reviewed.
Frequently asked questions
Should I do 100 APUSH questions a day?
No for most students. That volume usually prevents source/rubric review and crowds out writing.
How many full practice exams?
Use a small number as spaced checkpoints because review is expensive. Section-level practice can answer most diagnostic questions.
Do chapter quizzes count?
They can support content, but check whether they use AP-style stimuli and reasoning. Add official-format work when they do not.
The right number is the amount you can answer, analyze, and transfer while covering MCQ, SAQ, and essay skills. Start with the weekly range, adjust from the last error pattern, and count reviewed decisions—not pages of completed bubbles.