RRoadReady TAIWAN
RESEARCH BRIEF · UPDATED 16 AUGUST 2026

What is really in Taiwan’s new automobile written test?

The hazard-video percentage is official. Most other topic percentages are not. Here is the line between confirmed rules, measured bank composition, and useful—but non-guaranteed—estimates.

The short answer: since 30 June 2026, the ordinary automobile theory test contains 50 multiple-choice questions. Five are hazard-perception videos. The remaining 45 are selected from the published theory and image bank. The Highway Bureau has not published a rule saying signs must be 30%, 25%, or any other fixed share.

01

The format that is officially confirmed

The Highway Bureau’s reform announcement confirms 50 questions, five hazard-video questions, and a 1,100-item published bank. The authority’s simulator states that car questions are randomly selected, each is worth two points, and 85 points is the passing standard.

50questions
2points each
85official pass mark
43correct answers needed

Because scores move in two-point steps, 42 correct gives 84 and fails. You need at least 43 correct, producing 86 points. Seven mistakes can pass; eight cannot.

Highway Bureau reform announcement ↗Official random simulator ↗
02

The published bank has a measurable shape

The English static bank contains 1,090 questions. A separate file contains ten automobile hazard videos, bringing the official total to 1,100. The static PDF prints an exact total at the end of each of its three sections.

Published sectionQuestionsStatic-bank shareProportional expectation*
Correct concepts and ethics42438.9%17–18 of 50
Proactive yielding culture23021.1%9–10 of 50
Safe driving ability43640.0%18 of 50
Hazard videos10 available5 of 50 · confirmed
*The first three estimates assume the 45 non-video questions are sampled proportionally. They are a planning model, not an authority-guaranteed blueprint.
03

Are signs really 30%?

No current official source we found promises that. To estimate the likely exposure, RoadReady mapped every embedded image in all 86 pages to its question number, removed duplicates, and separated direct sign, marking, signal, and hand-signal recognition from other diagrams.

101direct visual-control recognition questions
9.3%of the 1,090-question static bank
≈4.2per test under proportional sampling

That is roughly 8.3% of a 50-question paper—not 30%—before counting text-only questions that discuss sign rules. The bank also contains 192 questions with at least one embedded image, or 17.6% of static questions.

Do not treat the estimate as a quota. Random generation can produce more or fewer visual questions, and the real examination system may use internal balancing rules that are not publicly documented.

04

A rational study allocation

For broad preparation, use the published bank proportions as your baseline, then deliberately overlearn the small hazard-video bank because five of its ten items appear in every test.

  1. 36%Safe driving ability
  2. 35%Correct concepts and ethics
  3. 19%Proactive yielding culture
  4. 10%Hazard videos

Give signs a dedicated 10–15% revision block within the first three categories. Learn shape, border or colour, central symbol, and exact official meaning—in that order.

05

What this analysis cannot tell you

The authority does not publish its production randomisation algorithm, per-topic quotas beyond the five videos, or frequency data from real candidate papers. The online simulator explicitly describes itself as practice and may not fully represent the actual examination. Anyone claiming a guaranteed “sign percentage” should be able to show a current Highway Bureau source.

Method and disclaimer

Counts were computed from the Highway Bureau’s English Year 115 PDFs available in August 2026. RoadReady is independent, does not guarantee test composition or a passing result, and is not a substitute for current authority instructions.