A father came to see me with two printed score reports on the table between us. One belonged to his son, who had taken the TOEFL in November. The other belonged to his nephew, who had taken it in March. The first said 103. The second said 5.0. He wanted to know which boy had done better.

It is a fair question. The honest answer is that the two numbers are not measuring the same thing, because the two boys did not sit the same examination.

On 21 January 2026, ETS replaced the TOEFL iBT with a substantially different test. Most of what has been written since then concentrates on the pattern: the exam is shorter, two sections now adapt, the scores look unfamiliar. All of that is accurate, and I will go through it carefully. But the pattern is the least interesting part of this story.

The question worth asking is why a company would rebuild a product that was profitable, accepted by institutions in more than 160 countries, and carrying no obvious defect. Once you understand the answer to that, nothing about the new format looks arbitrary any more.

01What changed, and when

The updated test took effect on 21 January 2026. If you sat the exam on or after that date, you took the new version. If you sat it before, your result is recorded under the old structure and remains valid for 2 years from your test date.

Here is the comparison, taken from ETS's own published structure.

Source: ETS, Test Content and Structure of the TOEFL iBT Test.

Three smaller changes matter more than they appear. Score reports issued between January 2026 and January 2028 include three elements: the CEFR level, the 1–6 band scores, and a comparable overall score on the old 0–120 scale. From January 2028, only the band scale survives. And ETS has retired paper score reports entirely.

02Before the pattern, the purpose

Students almost never ask why an English test exists at all, and it is the one question that explains everything else.

Put yourself on the other side of the desk. An admissions officer at a university in Illinois is handling 400 applications. One is from Warangal, one from Lagos, one from Hanoi, one from Bogotá. Each carries a school-leaving certificate marked by different examiners, against different standards, in a language the officer cannot verify. The officer has to answer one narrow question: can this person sit in a seminar next August, follow a lecture at natural speed, argue a point in a group project, write to a professor without causing confusion — and do all of it while also learning the subject?

A proficiency test exists to answer that question and nothing else. It is not a measure of intelligence or worth. Students who test poorly and go on to thrive are real, and so are students who test well and struggle once term starts. But a university cannot know any of that from a file, and the cost of guessing wrongly falls almost entirely on the student.

03Why a company rewrites a test that was working

Rebuilding a high-stakes examination is expensive and risky. Every university that accepts your scores has to be persuaded to re-anchor its requirements. No organisation does this casually. There are five reasons, and only some of them are educational.

First, English itself moved. The old TOEFL was built around a lecture hall, a long printed passage, a professor speaking uninterrupted for five minutes. A student today reads a course announcement on a learning-management portal, emails a professor about a missed deadline, posts on a discussion board, and negotiates a lab schedule over messages. A test that ignores all of this measures one register of English very well, and the rest not at all.

Second, coaching. When a test becomes predictable, preparation quietly stops being about language and starts being about templates — a memorised essay skeleton with three invented examples. The score rises; the underlying English does not. A redesign is partly an admission that the previous version had been solved.

Third, measurement. In a fixed test, most questions are wasted on any given candidate. An adaptive test concentrates questions near your actual level, where each answer carries the most information — which is why the new exam is shorter without being less precise.

Fourth, interpretability. The 0–120 scale was internally consistent but externally meaningless. The CEFR scale (A1–C2) is already the reference point for English teaching across Europe and, increasingly, in Indian classrooms. A 5 in any section now corresponds to C1 in that skill, which was not true of the old section scores.

Fifth, commercial. The Duolingo English Test can be taken from home in about an hour for roughly a third of the TOEFL fee, and is now accepted by a very large number of institutions. IELTS remains entrenched across the Commonwealth. At roughly ₹16,900–₹18,000 in India and a two-hour centre appointment, the TOEFL was losing on price and convenience. A shorter, faster, home-friendly test with a three-day turnaround is as much a competitive response as an educational one.

A commercial motive does not automatically make a change wrong. It does mean you should judge the result on evidence rather than the announcement.

If your primary constraint is cost, the fee gap between TOEFL, IELTS and the Duolingo English Test is real and worth weighing against how many institutions on your shortlist accept each one — cheaper is only a saving if the test is actually accepted where you're applying.

04The new examination, section by section

The order is fixed: Reading, then Listening, then Writing, then Speaking, with no scheduled break.

Reading now carries three task types. Complete the Words asks you to restore missing letters inside an academic paragraph. Read in Daily Life presents short practical texts — an email, a notice, a menu. Read an Academic Passage retains recognisable academic comprehension, but the passages are far shorter than the 700-word extracts they replace.

Listening has four task types: a short-response task, a conversation, an announcement, and an academic talk. The talks are considerably shorter than the old lectures, the content mixes academic and campus contexts, and the audio plays only once — the old replay questions are gone.

Writing has changed more than most students realise. Both tasks that demanded sustained argument have been removed: the 20-minute integrated essay and the independent opinion essay no longer exist. In their place: Build a Sentence (order jumbled words into a grammatical sentence), Write an Email, and Write for an Academic Discussion, which carries over from the previous format.

Speaking has changed most of all. All four earlier tasks are gone. What remains is Listen and Repeat — hear a sentence and reproduce it — and Take an Interview, a sequence of short spoken questions. There is no preparation time anywhere in the section, which now runs about eight minutes rather than sixteen.

Every task that requires you to combine sources, sustain an argument over several minutes, or synthesise a lecture with a reading has been removed. Whether that is a gain or a loss is the substance of section 07.

05The part almost everyone reads wrongly

The scoring change is not cosmetic, and the arithmetic has consequences that students discover too late.

Under the old system, your total was the sum of four section scores out of 30. Under the new system, your overall band is the average of the four section bands, rounded to the nearest half band. ETS gives the worked case: an average of 5.125 is reported as 5, and an average of 5.25 is reported as 5.5.

Two things follow. The first is compression: a band of 5 corresponds to a range of roughly 95–106 on the old scale, so a student who once stood out at 106 and one who scraped 95 now appear identical on the headline number. The second follows from the first — because the overall band can no longer distinguish strong candidates, ETS is explicitly advising institutions to set minimum section requirements rather than relying on the overall band alone.

ETS has published its recommended re-anchoring for institutions:

Source: ETS, TOEFL Score Scale Update — A Guide for Institutions.

There is a subtlety here that benefits students. Some universities that previously demanded 100 have been tempted to set 5.5 instead of 5, reasoning that a band of 5 begins at 95. ETS has advised against this, since raising the bar to 5.5 would exclude many candidates who would have comfortably cleared 100 on the old scale. If a university on your list has quietly set 5.5 where it used to ask for 100, that is worth raising with the admissions office directly.

06So is the new test easier?

Shorter is not the same as easier, and I would be cautious with anyone who tells you otherwise.

Some individual items are genuinely simpler — the daily-life reading and listening material demands less than the dense academic passages it partly replaces. Students who found the old integrated tasks punishing will feel real relief.

Against that, the adaptive design means your performance early in Reading and Listening determines which second module you receive, and prep specialists report that this routing places a practical ceiling on the band that section can reach. ETS has not published an explicit statement to that effect, so treat it as a well-founded inference rather than an official rule. Either way: the opening minutes of those sections now carry more weight than they used to, and casual early errors are expensive. Speaking gives no preparation time, Listening plays once, and the Speaking and Writing rubrics appear tightened — a top band now requires a genuinely strong response, not a merely competent one.

07The criticism that deserves a hearing

I would not be doing my job if I presented only the official account.

A research report published in 2026 by Ren and colleagues argues that the revised TOEFL represents a substantial shift in what the test actually measures, given the reduced length and complexity of its texts, the narrower range of task types, and what the authors consider an increased susceptibility to coaching. The report cautions specifically against relying on older concordance tables to treat new and old scores as equivalent.

That argument should be read with one fact in the open: the report was published by IELTS, a direct commercial competitor. That does not make it wrong, but it does mean it is not disinterested.

The underlying concern is nonetheless serious, and I share part of it. A test that no longer asks you to read a passage, listen to a lecture, and reconcile the two has stopped measuring something that university study genuinely demands. Several of the new task types also resemble those on the Duolingo English Test and ETS's own shorter TOEFL Essentials. Whether the new exam predicts academic success as well as the old one is an empirical question, and the evidence will take some years to accumulate.

My own position is that the redesign is defensible on measurement grounds and genuinely useful on interpretability, while the removal of every integrated task is a real loss that the convenience does not fully offset. You do not have to resolve that debate. You have to apply next year.

08What you should actually do

For students sitting the exam from here onwards, the preparation has to change, not merely intensify.

• Discard old practice material. Books and mock tests built for the pre-2026 pattern train you for tasks that no longer exist, at a pace that no longer applies.
• Use official ETS sample tests for the new format. For an adaptive exam, authentic material matters more than volume.
• Practise switching registers quickly. In ninety minutes you move from an academic paragraph to a notice to a conversation to a discussion post — the difficulty now lies in the switching as much as any single task.
• Work on the email task deliberately. Indian English tends towards the formal and the ornate; this task rewards plain, direct, functional writing.
• Read aloud daily for Listen and Repeat. It rewards pronunciation, rhythm and short-term memory — trainable, and rarely practised.
• Treat grammar as a separate skill again. Build a Sentence tests word order directly, with no essay to hide behind.

On timing: if you already hold a score from before 21 January 2026, it remains valid for two years from your test date, and reports will carry a comparable 0–120 figure until January 2028. Do not retake the exam merely because the scale has changed. If you are applying for entry in 2027 or later, you will be on the new format — plan your preparation around it now, not the pattern a senior described last year.

Before you book, check the requirements on the specific program page, not the general international admissions page, and note whether the university has published its 1–6 requirements yet. Many have; some are still quoting only the old scale, and in those cases a short email to the admissions office is worth more than a guess.

† IMFS Insights Team. IMFS Admissions Intelligence Dataset (v2026.1). IMFS, June 2026. See the full IELTS Score Target analysis on IMFS Insights.

09A closing thought

The 2026 TOEFL is neither easier nor harder. It is measuring a somewhat different thing, for reasons that are partly scientific, partly practical and partly commercial — on a scale that says more about you and reveals your weakest skill more plainly.

For a student who actually uses English, that is good news. The gap between real ability and coached performance has narrowed, and it was always the students with genuine command who were disadvantaged when it had not. For a student planning to memorise their way through, the news is less comfortable — better to hear it now than in the examination hall.

What I would not do is panic, rush to sit the exam before understanding it, or accept a conversion chart as gospel. Find out precisely what your universities require on the new scale, prepare for the test that exists rather than the one described in a three-year-old book, and give yourself enough time to improve the skill rather than the technique.

Behind every one of these score reports is a young person deciding what to do with several years of their life, and a family deciding where to place a very considerable amount of its money. That is worth a little care.