OET Writing: The Letter, The Case Notes and The Six Criteria

Search for OET Writing and Google returns letter samples. Five of the eight related searches it offered on 27 September 2026 named a sample or a PDF, and the top related query in India on Google Trends was oet writing sample. What almost nothing on that first page tells you is how the letter is actually scored, even though Google’s own People Also Ask box carries the question. The answer is that five of the six criteria are marked out of 7 and one is marked out of 3, which changes what a study hour is worth. What follows is the scoring system, and what OET’s published score data shows about who loses marks in it. For the test as a whole, start with how the OET works.

One task, 45 minutes and a set of case notes you may reread throughout

The Writing sub-test contains a single task: a letter to another healthcare professional, or to someone involved in the patient’s care, written from case notes. The timing is split rather than continuous, and the longer of the two parts carries a permission most candidates do not know they have.

Sources: OET’s Writing preparation page for the word range, layout, marking and reading-time rows, and the OET Test Handbook 2026 for the timing, stimulus, task type and profession rows. Both read 27 September 2026.

The detail worth acting on is that the five minutes of reading time is not the only time you may spend with the case notes. OET states that you can consult the task and the case notes at any point during the 40 minutes, not just during the reading time. That turns the five minutes into planning time rather than your only chance to read, and it makes a mid-letter check against the notes a legitimate move rather than a lost minute.

Six criteria, two different scales, and a maximum raw score of 38

Both the published level descriptors and OET’s own writing guide set out the scales, and they are not uniform. Each criterion is assigned a score from 0 to 7, except Purpose, which is scored from 0 to 3. Five criteria on a 7-point scale and one on a 3-point scale give a maximum raw total of 38, and the six criteria are not worth the same.

Sources: scales from the OET Ultimate Writing Guide; descriptors from the Writing assessment criteria and level descriptors; each criterion name links to OET’s own guide post for it. Share percentages are our arithmetic on a 38-mark maximum, rounded, so the column sums to 98 rather than 100.

In its guide post on Conciseness and Clarity, OET separates two criteria that are easy to read as one. Content assesses the information you have included in your letter, while Conciseness and Clarity assesses the information you have omitted from it, a framing OET concedes sounds odd because it means being assessed on something that is not there. Those are 37 per cent of the raw maximum between them, and they pull in opposite directions: every sentence you add helps one and threatens the other.

Grade B is published as a band profile, and that profile works out near 71 per cent

OET publishes the band profile a Grade B performance typically reaches, and it is specific rather than an aggregate. Test-takers securing grade B will have achieved scores of 2 out of 3 for Purpose and 5 out of 7 for the remaining criteria. Add those and a B profile totals 27 of the 38 available marks, or 71.1 per cent.

OET states that a Grade B in Listening and in Reading typically needs at least 30 of 42 marks, which is 71.4 per cent. Two different mark schemes land within half a percentage point of each other, which is worth knowing mainly because it shows the Writing profile is not a softer target than the objective sub-tests. The two figures are not identical objects, since OET combines and converts Writing criterion scores rather than totalling them against 27, so read the match as a sanity check rather than as a shared cut score.

At band 5 the published descriptors read: content is appropriate to the intended reader and mostly addresses what is needed to continue care, with content from the case notes generally accurately represented; organisation shows occasional lapses in sub-sections and in the highlighting of key information; and minor slips in language generally do not interfere with meaning. Band 7 is the level requiring no important details missing and no irrelevant information included. A B asks for mostly right with minor slips, not for a clean letter.

The scale is also coarser than eight points suggests. In the published descriptors the odd bands carry the written criteria, while bands 6, 4 and 2 are defined only as performance sharing features of the bands either side, and band 0 as performance below band 1. Four described levels therefore sit under each 7-point criterion, at bands 7, 5, 3 and 1, with three transitional scores between them. Purpose is different again: its bands 3, 2, 1 and 0 are all described, with no transitional scores at all.

Purpose runs on a three-point scale, so one band there costs a third of the criterion

Purpose is 3 marks of 38, around 8 per cent of the raw maximum, and its scale is coarser than the others. One band on Purpose is a third of that criterion’s marks; one band on Language is a seventh of its. The published B profile shows the same asymmetry: 2 of 3 here against 5 of 7 elsewhere leaves two score points of slack above the B level on every 7-point criterion and one above Purpose. OET combines and converts the criterion scores rather than gating each one, so treat that as a description of where the scale is unforgiving rather than as a pass mark per criterion.

OET splits the criterion into two parts. The first is that the reason for the letter is immediately apparent, so the reader does not have to search for it. The second is that the reason is sufficiently expanded later in the document, often near the end. The band descriptors price those two parts precisely: band 3 requires the purpose to be immediately apparent and sufficiently expanded, while band 2 covers a purpose that is apparent but not sufficiently highlighted or expanded. A letter that states its reason clearly and never returns to it therefore lands on band 2, which is exactly the level the published B profile names. It has not failed the criterion. It has used the only spare point the scale allowed it.

OET’s guide post on Language draws the line this way: the other criteria ask whether you used language appropriately, while Language asks whether you used it accurately. Appropriateness is judged against the particular reader the task sets, which is why the same sentence can score differently depending on whether the letter goes to a specialist, a general practitioner or the patient’s family.

Two assessors mark each letter, and OET reports its reliability with a Rasch model

Writing and Speaking are the only sub-tests scored by assessors applying criteria rather than against a right-or-wrong key, and OET’s own statistics treat them differently for that reason. Two OET assessors use the official criteria on every letter. Listening and Reading are reported as dichotomously scored, while Writing and Speaking are treated as polytomously scored, with reliability reported using a Many-Facet Rasch Model marginal reliability estimate and classification figures at the C+ and B boundaries computed with the Livingston and Lewis procedure. Separating candidate ability from the severity of the individual assessor is what that family of models exists to do, though OET publishes the reliability estimate rather than any statement about how severity is handled in your score.

The published figures change how a borderline Writing result should be read, and they differ by mode.

Source: OET test performance 2025, the average figures for the combined 2025 administrations, reached from OET’s test statistics page. The method names above are the PDF’s; the landing page itself describes its reliability measures differently, so take the PDF as the source. The ranking column is our comparison across the four sub-tests in the same table.

Classification consistency, on the standard reading of the index rather than on any definition OET prints here, is the expected agreement of the grade decision at a boundary across parallel administrations, so 0.82 on computer means that decision would be expected to repeat in about 82 of 100 cases. Read the row below it before concluding anything about Writing generally: on paper the same figure is 0.86 and is the highest of the four sub-tests, so the softness at the B boundary belongs to one mode rather than to the skill. That makes the mode a scoring variable and not only an interface choice, which is worth settling alongside what the test looks like before booking.

Writing scores compress near the threshold: 36 points separate the strongest and weakest nationalities

OET publishes mean scaled scores and median grades by gender, nationality, first language and profession in its annual test statistics. We parsed the 2025 release for all 67 nationality groups it reports, and Writing behaves unlike the other three sub-tests.

Source: our analysis of OET candidate performance 2025, mean scaled score by nationality, all 67 groups it publishes. Means, ranges and counts are ours; the underlying scores are OET’s.

Across these 67 group means Writing is the lowest of the four sub-tests, at 340.9 against 377.3 for Speaking, though the same PDF’s gender table puts Reading lowest for female candidates, so this is a property of the national aggregate rather than of every cohort. It also has the narrowest spread, 36 points against 103 for Listening, so the distance between the strongest and weakest national cohorts is roughly a third of what it is in a receptive skill. And the ceiling is thin rather than absent. The strongest Writing cohorts, four of them tied at 354, sit four marks above the Grade B threshold, while the weakest Speaking cohort sits five marks above it at 355. Ten of the 67 groups reach 350 in Writing; all 67 clear it in Speaking. Writing’s best national average is barely better placed against the threshold than Speaking’s worst.

Compression is not the same thing as difficulty, and the grade data separates them. The median grade in Writing was B in all 67 nationality groups and in all 12 professions. Listening’s median was C+ in 11 of those nationalities and Reading’s in 13. The median candidate is graded B in Writing and may be graded below B in Reading. Nor can a strong Writing score carry a weak one elsewhere, and the reason is structural rather than statistical: OET states in the same document that it does not report overall scaled scores or grades, only a score for each individual sub-test.

Indian candidates score higher in Writing than in Reading or Listening, which inverts the usual assumption

In our classes the assumption arrives fully formed: Writing is where Indian healthcare professionals lose the test. OET’s 2025 figures say otherwise.

Extracted from OET candidate performance 2025, mean scaled score by nationality. The table shows six of the 67 published nationality groups.

For Indian candidates the weakest sub-test in 2025 was Reading at 313, with Listening at 322, and Writing sat 30 points above Reading and 21 above Listening, at 343. A plan that spends most of its hours on the letter is optimising the third-weakest skill. The counter-argument is about room to improve rather than about risk. OET’s C+ band runs 300 to 340 and B starts at 350, so a cohort mean of 313 in Reading sits inside C+ while 343 in Writing sits above it, yet Reading cohorts reach 407 and Writing cohorts stop at 354. Both things are true, and the resolution is that Writing needs targeted repair rather than bulk hours, which is the pattern described in our note on where Indian doctors slip in OET.

The bottom three rows point at something the same PDF states more cleanly in a different table. OET also reports scores by first language, and the candidates whose first language is English averaged 377 in Listening, 360 in Reading, 394 in Speaking and 345 in Writing. Writing is the only one of the four they did not clear. Reading nationality rows as a proxy for language would be sloppier, and OET warns in the same document that nationality may not correlate with proficiency in the languages spoken in that country. It attaches a caveat to the first-language table too, that a self-reported first language is not an indicator of proficiency in it. With both caveats in view the row still points one way: the criteria are marking a professional document convention, not fluency.

Your registered profession sets the task, and the 2025 means differ by 37 points across the twelve

Because the task is profession-specific, OET reports Writing separately for each of the 12 professions. Only one averaged at or above the Grade B threshold in 2025.

Extracted from OET candidate performance 2025, mean scaled score by profession. Shown here are six of the 12 published professions; across all twelve the Writing means run from 318 to 355.

A profession averaging 318 in Writing averaged 385 in Speaking, a 67-point gap inside the same cohort. Whatever is happening there is not general English ability, and it is a reminder that a profession-specific task rewards familiarity with that profession’s documents. Practising on nursing case notes when you are registered for Radiography trains the wrong genre.

Re-mark talk clusters on Writing at more than twenty times the rate of Speaking

We searched a cleaned corpus of 824 Reddit threads about the OET posted between 1 February and 18 September 2026 for what candidates raise after a Writing result. One subject separates Writing from the other three sub-tests, and it is re-marking.

Source: our analysis of 824 Reddit threads dated 1 February to 18 September 2026. Ninety-four of the 98 Writing threads, and 729 of the 824 in the corpus, come from one subreddit for candidates pursuing US certification, so the sample skews toward doctors. Threads enter the corpus by subreddit or OET keyword rather than by mentioning OET in the text. A sub-test counts when its name appears in the title, and a re-mark when the title or body matches re-mark or re-score; fifteen titles name more than one sub-test and are counted under each. Counts are ours and reproducible.

Speaking generates 172 threads to Writing’s 98 and is also marked by human assessors, yet it produces two re-mark discussions to Writing’s 27. Across the whole corpus, 58 of the 77 threads that mention a re-mark name Writing. Counted once per thread across the 27 Writing threads that raise a re-mark, the figures named most often are 300, 350 and 290. Those three are the bottom of C+, the bottom of B and the top of C, so the talk sits on the grade boundaries rather than on low scores generally. The count takes any three-digit figure in a thread, so it picks up scores from other sub-tests too and is a rough signal rather than a distribution.

Six facts from OET’s re-marking policy decide most of these cases. A request must be made within 72 hours of results being released, and cannot be processed if it falls more than 14 days after the official release date, in which case OET refunds the fee in full. The fee is AU$120 per sub-test outside the United States, or US$95 inside it, separate from the current fee for the test itself. The outcome takes three to six weeks, and it can increase your score, decrease it, or leave it unchanged. The fee is refunded only if your grade changes, not if your score moves within the same grade.

A computer result follows in as little as 48 hours and most within six calendar days, against three to six weeks for a re-mark, so a candidate who re-sits can often hold a fresh score before a re-mark outcome arrives. The booking lead time that decides whether re-sitting is possible at all is set out in booking your test date sets out. Re-marking makes sense when you believe the band judgement was wrong; retesting makes sense when you need a number by a deadline.

The word range is 180 to 200, and missing it carries no automatic penalty

On this point OET’s position is unambiguous, and it is routinely restated as its opposite in the preparation material we see. There is no automatic penalty for writing over or under the 180 to 200 word range. The consequence runs through the criteria instead: OET states that a significantly longer letter probably contains irrelevant material or is poorly organised, and a significantly shorter one probably misread the task or the case notes and omitted something important.

Which means the word count is a symptom, not an offence. A 260-word letter does not lose marks for being 260 words; it loses them on Conciseness and Clarity because the extra 60 words are probably information the reader did not need, and possibly on Organisation and Layout because the key facts are now harder to find. Counting words at the end rarely points at anything you can fix. Checking which sentences the named reader would skip points at the cut, and the individual slips that follow from getting this wrong are catalogued in our post on what loses marks in Writing.

No layout is prescribed, and OET accepts the several formats used in different local contexts provided the letter is clearly laid out and suits the task. Capital letters throughout are permitted if you are consistent and do not switch between cases.

Demand asks for samples, so the band scales go unpublished on page one

Across the twelve months to 27 September 2026 the term oet writing registered zero in 50 of 54 weeks in India on Google Trends, which is too sparse to support any claim about seasonality, and we make none. What the two related queries Google did surface show is intent: oet writing sample, and oet writing sample for nurses.

The first page of results largely matched that demand. On 27 September 2026 it carried letter samples, sample PDFs, a video carousel from two commercial preparation channels, a Reddit thread and one of OET’s own criterion posts, above an AI Overview and a People Also Ask box whose second question was how the Writing sub-test is scored. Criterion content does rank, then: what no page-one result carried was the band scales or the published Grade B profile that answer the question. A page that publishes the scales, the Grade B profile and the score distribution is not competing with the sample pages; it is answering the question they leave open.

Sample demand is still legitimate, and it should be met with marked work rather than unmarked letters. A letter you cannot score against the six criteria tells you only that you finished. Our worked nursing example is set out in a marked nursing letter, and the last-pass routine that catches band-level slips is in the self-edit checklist.

A criterion-shaped plan beats a letter-shaped one

Writing preparation is usually organised around producing letters, which is what the test asks for. The scoring is organised around six criteria on two scales, five of them judging appropriateness for one named reader and one judging accuracy. Those are different objects, and practice aimed at the first does not automatically improve the second.

Know which of the six you are losing, because the six are not worth the same and they do not fail the same way: a 7-point criterion gives you two spare points above the B level, Purpose gives you one. Know that a B is 2 of 3 and 5 of 7, so the target is mostly right with minor slips rather than a perfect letter. And get letters marked by someone who scores them against the published descriptors rather than reading them for fluency, which is what OET coaching for the score you need is built to do, online and from the Egmore centre in Chennai.

Frequently asked questions

How is the OET Writing sub-test scored?

Two OET assessors score your letter against six criteria. Purpose is scored from 0 to 3 and the other five, Content, Conciseness & Clarity, Genre & Style, Organisation & Layout and Language, are each scored from 0 to 7, giving a maximum raw score of 38. The criterion scores are combined and converted to the 0 to 500 scale.

How do I pass OET Writing?

OET publishes the profile a Grade B performance typically reaches: 2 out of 3 for Purpose and 5 out of 7 for each of the remaining five criteria. That is 27 of 38 marks, or about 71 per cent. At band 5 the descriptors ask for content that is appropriate to the intended reader and mostly addresses what is needed to continue care, occasional lapses in organisation, and minor language slips that do not interfere with meaning, so a B does not require a flawless letter.

How long is the OET Writing sub-test?

45 minutes in total, made up of 5 minutes of reading time and 40 minutes of writing time. OET states you may consult the task and the case notes at any point during the 40 minutes, so the reading time is for planning rather than your only opportunity to read the notes.

How many words should an OET letter be?

The task is designed for 180 to 200 words, and OET applies no automatic penalty for going over or under. A letter that is much longer probably contains irrelevant information or is poorly organised, and one that is much shorter probably misses required content. Either way the loss appears in the criterion scores rather than as a word-count deduction.

Is OET Writing the same for every profession?

No. Writing is profession-specific, with tasks tailored to your registered profession, so Nursing candidates complete Nursing tasks and Radiography candidates complete Radiography tasks. Listening and Reading are the same for everyone. In OET’s 2025 statistics the Writing means by profession ranged from 318 to 355.

What is the difference between Content and Conciseness & Clarity?

OET describes Content as assessing the information you have included in the letter and Conciseness & Clarity as assessing the information you have omitted. Together they account for about 37 per cent of the raw maximum, and they pull against each other: adding a sentence helps one and threatens the other.

Do spelling mistakes matter in OET Writing?

Spelling sits under Language, along with punctuation and grammar, and errors are reflected in your Language score. OET’s own guidance notes that spelling matters but that you do not need to be perfect to achieve a B, since band 5 allows minor slips that do not interfere with meaning.

Is there a required layout for the OET letter?

No. OET accepts the several formats used by health professionals in different local contexts, provided the letter is clearly laid out and appropriate to the task, and you should use the title and address details given in the task instructions. Writing entirely in capital letters is permitted if you stay consistent.

Should I request a re-mark of my OET Writing score?

It depends on whether you think the band judgement was wrong rather than on how close the score was. A request must be made within 72 hours of results being released and cannot be processed more than 14 days after the official release date. The fee is AU$120 per sub-test outside the United States, the outcome takes three to six weeks, and the score can go up, down or stay the same. The fee is refunded only if your grade changes.

Which sub-test do candidates find hardest, Writing or Reading?

On OET’s 2025 figures it depends on what you mean. Writing had the lowest mean scaled score of the four sub-tests across the 67 nationality groups OET publishes, at 340.9, and 57 of those groups averaged below 350. But the median grade in Writing was B in all 67 groups, while Reading’s median was C+ in 13 of them. Writing is where headroom is scarce; Reading is where the median candidate is more often graded below B.