Which 2026 TOEFL questions people actually get wrong
Full access from CA$13.99 for 7 days · one-time, CAD, no renewal
Guides tell you what each task asks. This page tells you which ones people miss. Every question answered on ExamStep is marked against the answer key and tagged with what it was testing, so the accuracy below is counted, not estimated.
Counted live from our database · last updated · How to cite
On this page
Key findings
- All 18 grammar structures in Build a Sentence come in under 105% correct; equative comparison is lowest at 0.0% (1 questions).
- Attitude is the hardest comprehension question type at 33.3% correct across 3 questions.
- Questions that retrieve what the text says sit well above that: specific detail 77.3%, main idea 76.5%.
- Complete the Words is the weakest reading task at 63.2% (5,160 questions).
- In Listening, a conversation scores 79.5% and an academic talk 57.7%.
19,904 questions, 4,423 sets, 194 learners, 14 September – 8 November 2026. Practice data, not official TOEFL results.
Method: what was counted
19,904 questions across 4,423 practice sets, answered by 194 people between 14 September – 8 November 2026, each marked against the answer key. Staff and test accounts are excluded. Percentages are of questions, not of sets, and every row carries the number of questions and the number of people behind it — read the thin rows as a direction and the fat ones as a number.
This is early data about a test that only changed in 2026, over 56 days so far. It says what our practice material found; it is not a study of the official test, and ETS has nothing to do with it. The figures are counted live from our database and refreshed every five minutes, so the period, the counts and every percentage move together as more people answer; the date above says when they were last counted.
- Included: every scored TOEFL Reading, Listening and Build a Sentence set; each question marked right or wrong against the answer key and tagged with the task, structure or question type it tests.
- Excluded: staff accounts and our own test accounts; sets that were never scored.
- Window and sample: 14 September – 8 November 2026 — 19,904 questions, 4,423 sets, 194 learners.
- Other exams: ExamStep also runs CELPIP practice, but a skill is published here only once it has at least 30 learners, and no CELPIP skill has reached that yet, so CELPIP is not included.
- The raw figures are open as JSON at /api/public/toefl-question-data.
The new task is the hard one
Build a Sentence asks you to order fragments into a grammatical sentence. It is new in 2026, which means almost nobody arrives having practised it — and it is the only place on the test where a single grammar rule decides the mark outright. Every structure it put under test came in below 105%.
Equative comparison
0.0%1 questions · 1 person
Multiplier comparison
0.0%1 questions · 1 person
Negative adverb inversion
0.0%1 questions · 1 person
Passive voice
58.5%463 questions · 100 people
Embedded questions
62.0%902 questions · 158 people
Infinitive complements
66.4%250 questions · 100 people
Embedded whether / if clauses
66.7%459 questions · 157 people
Word order
68.7%508 questions · 97 people
Future going to location
100.0%1 questions · 1 person
Future passive condition
100.0%2 questions · 2 people
Future perfect
100.0%1 questions · 1 person
Modal perfect deduction
100.0%2 questions · 2 people
Not only inversion
100.0%1 questions · 1 person
Past simple object
100.0%2 questions · 2 people
Present continuous future
100.0%1 questions · 1 person
Present perfect infinitive
100.0%1 questions · 1 person
Relative clause defining
100.0%2 questions · 2 people
Reported speech
100.0%2 questions · 2 people
Equative comparison at 0.0% is worse than a coin toss, and multiplier comparison at 0.0% is worse than a coin toss. These 18 rows rest on 1 to 158 people, so treat them as a strong hint rather than a settled fact; they are also the rows we expect to move most as more people sit the task.
What a question asks decides how often it is missed
Across Reading and Listening together, accuracy tracks what the question demands rather than which section it sits in. Retrieving something the text says is comfortable. Working out how the text is built, or what it implies, is not.
Attitude
33.3%3 questions · 2 people
Organization
33.3%3 questions · 2 people
Sentence insertion
44.5%510 questions · 166 people · where a given sentence belongs in the passage
Sarcasm
50.0%2 questions · 2 people
Inference
60.5%613 questions · 174 people · what follows from the text but is never stated
Rhetorical purpose
63.0%1,238 questions · 173 people · why the writer or speaker said it
Vocabulary in context
63.9%5,643 questions · 167 people · the word the sentence needs
Pragmatics
69.0%2,332 questions · 140 people · what the speaker means, not what they said
Morphology
75.0%4 questions · 2 people
Main idea
76.5%893 questions · 174 people · what the whole thing is about
Specific detail
77.3%5,567 questions · 174 people · a fact stated in the text
Cohesion
80.0%10 questions · 2 people
Idiomatic
90.0%10 questions · 2 people
Cause and effect
92.5%372 questions · 167 people · which thing led to which
Grammar
93.3%15 questions · 2 people
Purpose
94.3%53 questions · 11 people
Function
100.0%13 questions · 2 people
Implied meaning
100.0%5 questions · 2 people
Implied refusal
100.0%1 questions · 1 person
Indirect question
100.0%1 questions · 1 person
Indirect request
100.0%6 questions · 2 people
Next action
100.0%6 questions · 2 people
Reference
100.0%2 questions · 2 people
Understatement
100.0%2 questions · 2 people
Attitude is the hardest question type on the test at 33.3%. Organization follows at 33.3%. At the other end, specific detail (77.3%) and main idea (76.5%) are right most of the time.
The practical reading of this: time spent re-reading for facts is time spent on the questions you were going to get anyway.
Accuracy by task
The same pattern one level up. What separates an easy task from a hard one is register: everyday material — conversations, notices, messages — scores above academic material.
Reading
Complete the Words
63.2%5,160 questions · 167 people
2,570 questions · 167 people
Read in Daily Life
93.0%2,060 questions · 167 people · sets of 2–3 questions
Read in Daily Life at 93.0% is worth understanding rather than celebrating: these are notices, messages and schedules in sets of two or three questions, and 668 of 773 sets were perfect. Complete the Words is the opposite — 30 of Reading's 50 items in the official blueprint, and our weakest reading task at 63.2%. The largest task on the section is also the one people miss most.
Listening
2,576 questions · 141 people
860 questions · 141 people
Choose a Response
69.3%2,362 questions · 141 people
1,716 questions · 141 people
A conversation between two people is comfortable at 79.5%. The same listener drops 22 points on an academic talk. Nothing about the audio is faster; what changes is that the talk carries an argument, and the questions ask about the argument.
Questions about this data
What is the hardest part of the 2026 TOEFL?
On our own marking, Build a Sentence. Every one of the 18 grammar structures it tests comes in under 105% correct — equative comparison at 0.0% and multiplier comparison at 0.0% — while ordinary comprehension questions such as specific detail (77.3%) and main idea (76.5%) sit well above. It is also the newest task on the test, so almost nobody has practised it.
Which TOEFL question type do people get wrong most often?
Attitude — this question type — at 33.3% correct across 3 questions.
Is TOEFL Listening harder than Reading?
They are closer than people expect, and the difficulty sits inside the task rather than the skill. Listen to an Academic Talk is our weakest listening task at 57.7%, Complete the Words our weakest reading task at 63.2%. What separates easy from hard is register: everyday material (a conversation 79.5%, daily-life reading 93.0%) scores above academic material (an academic talk 57.7%, Complete the Words 63.2%).
How much data is this based on?
19,904 questions in 4,423 practice sets, answered by 194 people between 14 September – 8 November 2026, marked against the answer key. Staff and test accounts are excluded. The figures are counted live and were last updated 8 November 2026, 05:50 Pacific time. It is early data on a test that only changed in 2026, and the page says on every row how many questions and how many people stand behind it.
Accuracy figures are counted live from questions answered on ExamStep and marked against the answer key, staff and test accounts excluded, for the period named above (last counted 8 November 2026, 05:50 Pacific time). Task names, item counts and the section blueprint are ETS, TOEFL iBT Test: 2026 Update — Test Blueprint and Specifications, checked 5 September 2026. No question, passage or recording is reproduced. ExamStep is an independent practice resource and is not affiliated with ETS. TOEFL® and TOEFL iBT® are registered trademarks of Educational Testing Service. Anyone is welcome to cite these figures with a link to this page.
How to cite this data
Please do. The figures on this page are published under CC BY 4.0, which asks only for attribution. Because they are live, give the date you read them. A line you can paste:
ExamStep (2026). Which 2026 TOEFL questions people actually get wrong. 19,904 questions answered 14 September – 8 November 2026; figures as of 8 November 2026, 05:50 Pacific time. https://examstep.ca/toefl/question-data
- URL
- https://examstep.ca/toefl/question-data
- Last updated
- 8 November 2026, 05:50 Pacific time
- Licence
- CC BY 4.0 — credit “ExamStep”
If you are writing something and want a figure cut a different way — by task, by question type, or over a different window — ask at support@examstep.ca. We would rather you had the real number than an estimate.
In this series: the whole 2026 format · what changed · Reading · Listening · Writing · Speaking · the 1–6 score scale.
Free reading practice, with answers
A complete worked example for every reading task type — the prompt, every question with its answer, and why the other options fail. No account needed.