The revised TOEFL iBT launched on 21 January 2026 and is reported on a 1–6 band scale. ETS has published the task list, the item counts, the timing and the scale. It has not published several of the things test takers most want to know — and the vacuum is filled by confident, mutually contradictory claims. This page lists each gap, records what other sites assert in its place, and says plainly what we do and do not claim.
By the LingoLeap TOEFL Content Team · Last reviewed:
As of 20 September 2026, ETS has not published: a raw-correct-to-band table for Reading or Listening; how a pre-21-January (0–120) and a post-21-January (1–6) sitting combine into one MyBest; how long a score hold takes; whether multistage routing limits the band you can reach; or the scoring model behind Listen and Repeat. Everything below marked "asserted" is somebody's inference, including ours where we say so.
What ETS publishes
Scores are reported 1–6 in half-point steps. The overall score is the average of the four section scores, rounded to the nearest half band — ETS's own worked examples are that an average of 5.125 gives 5 and an average of 5.25 gives 5.5. That is in the ETS 2025–2026 Information Bulletin, quoted in full on our score-release guide.
What ETS has not published
No mapping from the number of questions you answered correctly to a section band, for Reading or for Listening. ETS's 2026 content pages describe task names, item counts and timing, and stop there.
What is asserted elsewhere
estudyme publishes a complete table (6 = 44–47 correct out of 47, 5.0 = 36–39, 4.5 = 32–35) with no source given. toeflresources publishes a conversion table that is the pre-2026 0–30 scale. Recorded in our SERP audit of 16 September 2026.
Where we stand
Our own score calculator produces an estimate and labels it as an estimate. There is no official table to check any estimate against — ours included. Treat any unsourced raw-score table, ours or anyone's, as a guess.
What ETS publishes
MyBest is your best score for each section, combined across all valid TOEFL iBT scores from the last two years. From 21 January 2026 score reports use the 1–6 scale, and for a two-year transition a comparable 0–120 score is printed alongside. Whether MyBest is accepted at all is each institution's decision — the ETS FAQ says to check with the institutions you are applying to.
What ETS has not published
How a test taken before 21 January 2026 (reported 0–120) and one taken after (reported 1–6) combine into a single MyBest. The ETS pages we checked do not state it.
What is asserted elsewhere
Several prep and study-abroad sites describe a mixed-scale MyBest as if the rule were settled, generally by converting the old score and taking the higher number. None of them cites an ETS page for the conversion step.
Where we stand
We do not state a rule. If you hold sittings on both scales, go by what your actual score report shows, and show the report itself to the admissions office if there is any doubt. Our dedicated MyBest guide works through the parts that are documented.
What ETS publishes
Under "Holding/canceling scores", ETS may hold scores for investigation where it has information it considers sufficient to indicate activity affecting score validity. While scores are held you cannot send score reports to any institution until the investigation is complete.
What ETS has not published
Any duration. ETS publishes no expected or maximum length for a hold investigation.
What is asserted elsewhere
Forum answers and prep blogs circulate figures from "a few days" to "six to eight weeks". A related and more common error: the "4–8 days" score-release figure that still appears widely is ETS's line for tests taken *before* 21 January 2026 and for TOEFL iBT Australia. For the revised test ETS says scores are posted approximately 3 days after the test, and its November 2025 update notice puts it as scores delivered within 72 hours or less.
Where we stand
We do not invent a number. If a hold is confirmed, ask ETS customer service for a case reference and an expected timeline, and tell the universities you are applying to, because score sending is blocked while the hold stands.
What ETS publishes
Reading and Listening use multistage adaptive testing: a router module routes you into a lower or upper second-stage module. Writing and Speaking are not adaptive. That much is in the April 2026 ETS blueprint and is set out on our 2026 format and multistage strategy guides.
What ETS has not published
Whether being routed into the lower second-stage module bounds the section band you can still achieve, and by how much. ETS has not addressed it.
What is asserted elsewhere
Directly contradictory. leapscholar and college-council state that a perfect score on an easier Stage 2 caps you below 5.0. toeflresources calls the score-ceiling claim unverified test-prep speculation. Both were live in the same SERP on 16 September 2026.
Where we stand
Unverified, and we say so rather than picking a side. Our own mock test implements two-stage routing, and anything we publish about how routing behaves is a statement about our engine, not about ETS's.
What ETS publishes
That the task exists, where it sits in the Speaking section, and its timing and item counts.
What ETS has not published
The scoring model. ETS has not published what the automated scorer weighs on this task, or how a response maps onto the 1–6 band.
What is asserted elsewhere
prepdrills states that on the interview task the curve peaks at 3.0 and only 16% reach 4.0, against 75% at 4.0 or above on Listen and Repeat. No source is given for either figure. Elsewhere, bettertoeflscores states that SpeechRater and e-rater flag template phrases and that templated language costs points, and Study.com that rehearsed templates actively hurt scores; ETS has published no such rule.
Where we stand
Where we have measured something ourselves we publish it as ours, with the sample size, in our research section. We do not present our measurements as ETS policy, and we do not repeat the percentage claims above.
Everything above follows from four checks you can run on any TOEFL 2026 page, including this one.
Does it name the document?
"ETS says" is not a source. The 2025–2026 Information Bulletin, a dated ETS update notice and the April 2026 blueprint are sources. If a page gives a precise number and no document, the number came from somewhere else.
Does it distinguish before and after 21 January 2026?
Most wrong TOEFL figures on the web today are correct figures for the old test. The 0–120 scale, the 4–8 day score wait and every pre-2026 conversion table are all still true of tests taken before that date and false of tests taken after it.
Does it say what it does not know?
A page that answers all five questions above with confidence is not better informed than one that answers none of them. It is filling gaps ETS has left, and it should say so.
Does it separate its own data from ETS's?
A prep provider's own scored responses are real evidence about that provider's population. They are not evidence about ETS's scoring. Both are useful; conflating them is not.
A full-length mock test built to the 2026 task list and the 1–6 scale, with per-task feedback.
Take a free mock test