TOEFL still prints a score out of 120. It is no longer your score

Each of the four sections — Reading, Listening, Speaking, Writing — is now reported on a 1–6 scale in half-point increments: 1, 1.5, 2, 2.5, 3, 3.5, 4, 4.5, 5, 5.5, 6. Eleven possible values per section.

The overall result is the average of all four sections, rounded to the nearest half band. So the total is no longer a sum. The old 0–120 was four section scores of 0–30 added together; the new overall is four bands averaged. That is a different instrument, not a relabelling of the old one.

This puts TOEFL structurally alongside IELTS, documented on this site, which also averages four section bands and rounds to the nearest half band. The difference is what happens at the midpoint. IELTS publishes its tie-break and both stated cases resolve upward, with worked examples. The ETS pages read here do not state a tie-break rule — they say the average is rounded to the nearest half band and stop. So this guide does not assert how TOEFL breaks an exact tie in either direction. If your average lands precisely between two bands, that is a question for ETS, and it is worth asking, because on an eleven-point scale a half band is a large fraction of the whole range.

The transition is dated at both ends. The 0–120 comparable score appears on reports for two years, and the China-facing announcement is explicit about when it goes: not until January 2028 do reports show the 1–6 scale alone.

ETS publishes the alignment between the new bands and the old scale. Set out as closed ranges, it looks like this:

6 — C2 — 114–120. 5.5 — C1 — 107–113. 5 — C1 — 95–106. 4.5 — B2 — 86–94. 4 — B2 — 72–85. 3.5 — B1 — 58–71. 3 — B1 — 44–57. 2.5 — A2 — 34–43. 2 — A2 — 24–33. 1.5 — A1 — 12–23. 1 — A1 — 0–11.

Now combine that table with the sentence about what gets printed. The comparable 0–120 score is the midpoint of the corresponding total range. This arithmetic is ours. A candidate placed in band 4 sits somewhere in 72–85, and the midpoint of 72–85 is 78.5. So a candidate who would once have scored 85 and a candidate who would once have scored 72 now receive the same comparable 0–120 number. Thirteen points of difference, gone — not because the test measured them as equal, but because the report prints the middle of the box they were both put in.

The same collapse happens in every band. Band 3 spans 44–57 and reports around 50.5. Band 5 spans 95–106 and reports around 100.5. The familiar number survived the change; what it is a number of did not.

This matters most to anyone comparing an old score with a new one. A 0–120 figure issued before 21 January 2026 is a measurement. A 0–120 figure issued after it is a band label wearing the old number's clothes. They are not the same kind of object, they look identical on paper, and nothing in the format tells the reader which one they are holding.

This section is ours, and it is arithmetic on ETS's own published table. Take the width of each band on the old scale:

Band 3 — 14 points. Band 3.5 — 14. Band 4 — 14. Band 1 — 12. Band 1.5 — 12. Band 5 — 12. Band 2 — 10. Band 2.5 — 10. Band 4.5 — 9. Band 5.5 — 7. Band 6 — 7.

The widest bands are 14 points and the narrowest are 7. The scale is therefore twice as precise at C1 and C2 as it is at B1 and B2 — and the three widest bands in the set, 3, 3.5 and 4, cover the stretch from 44 to 85, which is where the largest share of candidates and a great many admissions thresholds live.

The practical consequence is worth stating plainly, and it is also ours. Institutional requirements were written against the old scale in round numbers — 80, 90, 100. Band 4 runs from 72 to 85, so it straddles a threshold of 80: an institution that used to require 80 and now accepts band 4 is accepting candidates who would have scored 72. One that moves up to band 4.5 is requiring 86. There is no band that means 80. Any institution translating an old integer requirement into the new scale has to move its bar, in one direction or the other, whether or not it intends to.

If you are applying, this is the question to ask and it is not rhetorical: which band do you require, and did your old number move up or down when you set it? The exam cannot answer it. Only the organisation reading your report can.

ETS publishes the shape of the current test: Reading — 50 items, 30 minutes. Listening — 47 items, 29 minutes. Writing — 12 items, 23 minutes. Speaking — 11 items, 8 minutes. Test time does not include directions. Please allow approximately two hours to complete the test.

The arithmetic is ours. The overall band is the average of the four sections, so each section carries exactly a quarter of the result. Speaking is 8 of those 90 minutes — under nine per cent of your time on the test, and twenty-five per cent of the outcome. Put the other way: one minute of TOEFL speaking carries the weight of 3.75 minutes of reading.

That is the same finding this site published about the Turkish TYS on 9 September, where a minute of speaking is worth exactly four minutes of reading, and it lands here for a different reason. The TYS assigns 25 points to each of four unequal papers by design. TOEFL arrives at almost the identical ratio by averaging four sections that happen to take very different lengths of time. Two exams, two mechanisms, the same answer — and in both cases the section candidates spend the least time on is the one with the most leverage per minute.

The item counts point the same way. Fifty reading items in thirty minutes is about 36 seconds each; eleven speaking items in eight minutes is about 44 seconds each. But twelve writing items in twenty-three minutes is nearly two minutes each. The sections are not merely different lengths — they are different densities.

The task names published for the current test are not the ones most preparation material describes. Reading: Complete the Words, Read in Daily Life, Read an Academic Passage. Listening: Listen and Choose a Response, Listen to a Conversation, Listen to an Announcement, Listen to an Academic Talk. Writing: Build a Sentence, Write an Email, Write for an Academic Discussion. Speaking: Listen and Repeat, Take an Interview.

And there is one sentence on that page which changes how every number above should be read: As the test adapts, test time and items may vary.

The test adapts. The 50 / 47 / 12 / 11 item counts and the 30 / 29 / 23 / 8 minutes are therefore nominal rather than fixed — what you actually sit depends on how you are performing while you sit it. This site records that as ETS states it and does not go further: the pages read here do not explain the adaptive algorithm, at what granularity it adapts, or whether two candidates in the same room receive the same number of items.

Two consequences follow for a reader planning their preparation. Practice material built around a fixed item count is describing a test that no longer runs that way. And Listen and Repeat and Take an Interview are not the independent-and-integrated speaking tasks that older guides drill. Check the task names against the current page before buying anything built to prepare you for them.

TOEFL iBT scores are valid for 2 years from the test date. ETS will not report or send scores after 2 years.

That is a hard stop, and it is worth setting against the other English exam documented on this site, because the two take opposite positions on the same question. IELTS does not expire a result at all: its partners recommend a maximum of two years to the organisations reading it, and state that an older Test Report Form should be accepted if it comes with evidence the candidate has kept their English up. The decision is delegated to the reader. TOEFL does not delegate it. After two years there is nothing to delegate, because ETS will not send the score to anybody.

This comparison is ours, and it is the sharpest version of an axis this site has been building for two weeks. Some bodies expire the certificate (TOPIK at two years, SIELE at five, the Turkish TYS at two). Some never expire it and say so in one word. IELTS refuses to do either and hands the judgement to the institution. TOEFL is the only one here that expires the service rather than the document: your report is not declared invalid so much as made undeliverable.

MyBest scores are the counterweight, and they run on the same two-year clock. ETS combines your best section scores across all valid TOEFL results from the last two years into a single set. There is a wrinkle created by the scale change and it is stated: for tests taken on the 1–6 scale, older 0–120 results are converted into the current format; for tests taken before it, MyBest remains 0–120 only. So a candidate straddling January 2026 can hold two MyBest sets that are not expressed in the same units.

From 9 to 11 September 2026 this section said there is no global TOEFL price. It was wrong, and the sentence it quoted was the clue: Select your testing location to view the registration fee. Behind that selector sits one ETS file that lists a US-dollar fee for 212 countries and territories, read on 11 September 2026. The page adds that TOEFL iBT Prices exclude VAT or other applicable taxes, except for EU countries.

The spread is US$170 to US$475. The cheapest sittings are Nigeria and Ukraine at US$170, then India US$173, Egypt and Kazakhstan US$175. The dearest are Switzerland US$475, Israel US$425, Finland US$377, Norway and Sweden US$375. The most common price is US$255, charged in 50 countries. Germany is US$280, France US$282, Italy US$305, the Netherlands US$303, the United Kingdom US$260, the United States US$270, Japan US$195, Turkey US$185, China US$283.17.

The same test costs 2.8 times as much depending on where you sit it. The ratio is ours; both prices are ETS's. It is the same posture as Turkey's TYS, which publishes a price for fifty-seven countries, and the SIELE, whose booking page prices ninety-five — and it is the widest spread of the three: the SIELE runs 135 € to 175 €, a ratio of 1.3. All 212 TOEFL rows and all 95 SIELE rows are printed in the fees-by-country dataset. ETS also publishes, in US dollars and the same everywhere, what everything around the test costs.

Express registration (within 7 days) — US$49. Rescheduling — US$69. Reinstatement of cancelled scores — US$20. Additional score reports, per institution — US$29. Speaking or Writing section review — US$80. Speaking and Writing section review — US$160. Express scoring — US$129. Returned payment — US$30.

So the eight prices around the test are flat, in dollars, while the test itself is priced per country — and the arithmetic of that is uneven. A candidate in Ukraine pays US$170 to sit and US$80 to have one section reviewed: the review is nearly half the exam again. A candidate in Switzerland pays US$475 to sit and the same US$80: a sixth. The review costs the same everywhere; what it costs relative to the exam varies by country as much as the exam does.

One detail in that list deserves its own line, because it is a real limit rather than a price. Only Speaking and Writing can be reviewed. There is no review offered for Reading or Listening at any price. Compare IELTS, which lets you submit any section for a re-mark and refunds the fee if your band changes. TOEFL's review is confined to the two sections a human judges, and the pages read here do not state that the fee is refunded if your score changes — so this guide does not claim it is, in either direction. Ask before you pay.

One structural point that is easy to miss in the alignment table. The lowest band, 1, maps to A1, and its comparable range starts at 0.

This observation is ours. That means a TOEFL result of zero and a TOEFL result of eleven are the same band, and that band is labelled A1 — a level that describes someone who can introduce themselves and ask simple questions. The scale has no value that means "no measurable English", because its floor is a CEFR level rather than an absence of one.

IELTS resolves this differently and explicitly: its scale runs down to Band 0 — Did not attempt the test, a value that describes a non-event rather than an ability. TOEFL's floor is occupied. Whether that matters depends entirely on who is reading your report, but anyone treating band 1 as evidence of A1 ability should know that the band also contains everyone below it.

At the top the compression runs the other way. Every CEFR level in the mapping covers two bands — A1 is 1 and 1.5, B2 is 4 and 4.5, C1 is 5 and 5.5 — except C2, which is a single band, 6, starting at 114 of the old 120. It is the only level on the scale with nowhere above it and no half-step inside it.