Free TOEFL Practice Questions, by Task Type

A practice test measures. This drills.

If you have taken a section test and come out with a Reading band of 3.5, you know the score and not the cause. Reading is 50 items in 27 to 30 minutes across three different task types, and a 3.5 can be built out of weak Complete the Words work, a mishandled academic passage, or careless marks on short daily-life texts. Those three problems have nothing in common, and none of them gets solved by taking another full test.

If you do not yet know where your marks are going, start with a free TOEFL practice test and read the per-question results before you come back here. If you already know, pick that task type below and work its ladder. Each one is 100 questions arranged as 20 levels of 5, so a level is two to four minutes, short enough to start on a phone and long enough that the result means something.

This works because TOEFL task types are narrow. Build a Sentence is a limited set of syntax patterns. Choose the Response is a limited set of speaker intents. Complete the Words is word ordering rules plus vocabulary recall. None of them improve much from mixed practice, and all of them improve quickly from a few focused sessions on the one task.

The ten task types, and how many marks each one carries

Reading is 50 items in 27 to 30 minutes. Complete the Words is 30 of those items on its own, which makes it the largest single block anywhere on the test. Read in Daily Life is 5 to 15 items and Academic Passages is 5 to 15 items. Students who spend all their reading preparation on long academic passages are working on the smaller half of the section.

Listening is 47 items in 25 to 29 minutes: Choose the Response 15 to 19 items, Academic Talks 8 to 16, Conversations 10, Announcements 6 to 10. Every clip plays once. That single fact is why Listening rewards different practice from Reading. You are training recall and intent recognition against a clock you cannot pause, not comprehension.

Writing is 12 items in about 23 minutes, and 10 of those 12 are Build a Sentence. The other two are one email, 7 minutes for 130 to 140 words, and one academic discussion, 10 minutes. Most students put all of their writing preparation into the two pieces of continuous writing and none into the ten items that outnumber them.

Speaking is not drilled here. It is 11 items in about 8 minutes, seven Listen and Repeat sentences plus a four-question interview at 45 seconds each, and it needs a microphone and feedback on delivery rather than a right-or-wrong ladder. That lives in the AI Speaking Tutor. The full section-by-section inventory is on the 2026 test format page.

What one level is, and what clearing it means

A level is 5 questions and you need 3 of them to open the next one. Not 5, which turns a drill into a wall you stop coming back to, and not 1, which lets you reach level 20 having learned nothing. Five questions also gives a clean result: 0, 20, 40, 60, 80 or 100 percent, so you can tell a good level from a lucky one.

The 20 levels are grouped into four bands. Foundations, levels 1 to 5, is everyday language and short texts, and its job is to make the mechanics of the task automatic before the content gets hard. Building, levels 6 to 10, brings longer material and less familiar topics, which is where the task starts to cost you attention. Test Ready, levels 11 to 15, is real exam length at real exam difficulty. Beyond Test, levels 16 to 20, is past what the exam uses on purpose: denser text, faster audio, closer distractors.

Levels 1 to 15 sit inside the official benchmark for that specific task, so clearing level 15 means the task itself will not surprise you on test day. Levels 16 to 20 exist so that the real thing feels like a step down instead of a step up. Any level you have cleared can be replayed and only your best attempt counts, so going back to a level that beat you costs you nothing.

Every answer is explained after you check it, which is where the learning actually happens. Being told the right letter teaches nothing. Being told that the option you picked repeats a phrase from the passage while the correct one paraphrases it teaches you a trap family, and the trap families repeat all the way through the test.

What the 2026 format changed, and why task-by-task practice works now

ETS relaunched the TOEFL iBT on January 21, 2026. The test now runs about 90 minutes with no breaks, down from roughly two hours. Scores come back on a 1 to 6 band scale in 0.5 increments, aligned to CEFR levels, reported alongside the legacy 0 to 120 scale through 2028.

The change that matters most for how you prepare is what happened to the tasks. The integrated tasks, where you read and listened and then wrote or spoke about the same material, are gone. What replaced them is a set of independent, practical tasks: completing words inside a paragraph, reading an email or a group chat, choosing a spoken reply, assembling a sentence from chunks, writing an email. Those are separable skills, which is exactly what makes them drillable one at a time. The old format resisted this, because a weak integrated writing answer could have come from the reading, the listening or the writing and you could not tell which.

Reading and Listening are also adaptive now, at the module level rather than question by question. Everyone gets the same first module, and the second module comes back harder or easier depending on how the first one went. In practice that means early accuracy sets your ceiling, so the task types that fill the section, Complete the Words in Reading and Choose the Response in Listening, are the ones worth making automatic before test day.

How to find the task that is costing you the band

Take a section test and count your misses by task type, not by section. A Reading score of 28 out of 50 tells you almost nothing. The same 28 with 14 of the 22 misses sitting inside one task type tells you exactly what to do this week.

Three rules that hold up. If more than a third of your misses sit in one task type, that is your task, and everything else can wait. If the misses are spread evenly, start with the highest-volume type in that section, which is Complete the Words for Reading, Choose the Response for Listening and Build a Sentence for Writing, because that is simply where the marks are. And if you are missing listening questions across every task while understanding the words, the problem is not vocabulary, it is that you are still processing the previous sentence when the next one arrives, and Choose the Response is the fastest way to train that out.

Then work that one ladder until you clear Test Ready, and take another section test. Re-measuring after a focused week is how you find out whether the drilling moved anything. Drilling for a month without re-measuring is how people end up with a lot of practice and the same band.

What a band means for each skill, and what moves it

The 1 to 6 scale runs in half-point steps and maps to CEFR, and band 4.0 and above is competitive for most universities. Because the steps are half a point, the distance between the band you have and the band you need is usually a small number of items rather than a rebuild of your English. That is the case for drilling: a handful of items, concentrated in one or two task types.

In Reading the gap between a 3 and a 4 is usually accuracy under time rather than difficulty. A student who gets 90 percent right with unlimited time and 60 percent right in 30 minutes does not have a reading problem, they have a decision-speed problem, and the fix is volume on whichever task is eating the clock.

In Listening the gap is memory and intent, not vocabulary. Conversation and talk questions ask what a speaker meant, implied or was trying to do, which is why students who understood every word still miss them. One hearing with no replay is the only condition that trains it, and it is the condition every level here uses.

In Writing the two halves move at different speeds. Build a Sentence is pattern knowledge and can improve inside a week, because there is a limited stock of structures and you either know how a relative clause attaches or you do not. The email and the discussion move more slowly, because they turn on register, on backing a reason with a concrete example, and on engaging with what another person actually wrote.

A week of practice built out of these levels

Twenty minutes a day is enough, as long as the twenty minutes point at the right task. Four levels a day is 20 questions. Across six days that is 120 questions aimed at your weakest area, which is more focused practice than most students get from a month of full-length tests.

A workable week: two days on the highest-volume task in your weakest section, two days on the specific task type your results flagged, one day alternating between them, and one day on a timed section test to re-measure. Replay any level you scraped through on 3 out of 5 before moving up, because 3 out of 5 means two questions went wrong and nothing has been done about the reason.

One habit is worth building deliberately: read the explanation on the questions you got right, not only the ones you missed. A question you guessed correctly is a question you will miss next month. The explanation is the only thing that tells you whether you knew it or found it.