OET Speaking for Doctors 2026: Format and Practice

🦈 OET Speaking is a conversation, not a quiz — and conversations only get better by having them. SpeakShark gives you daily spoken-English practice with an AI teacher that plays the patient, corrects your pronunciation mid-sentence, and remembers the mistakes you keep repeating between sessions. Free tier, no card. Start a free speaking session →

OET Speaking for doctors is two profession-specific role plays of about five minutes each, with three minutes of preparation before each one, running roughly twenty minutes in total — you play the doctor, a trained interlocutor plays the patient or their carer, and at least two OET assessors mark the recording afterwards. For UK registration the bar is clear and published: the GMC requires the Medicine version of OET with at least grade B in speaking, listening, reading and writing, all in the same sitting. This guide covers the exact format, both scoring criteria sets with the numbers OET publishes, what the GMC and Australia's National Boards actually demand in 2026, and six original doctor role-play scenarios you can rehearse tonight.

In this guide: at a glance · what actually happens · how it is scored · what grade you need · 6 original role-play scenarios · your three preparation minutes · pronunciation traps · how we researched this · when this is the wrong prep · building a daily habit · FAQ

OET Speaking for doctors at a glance

Two things in that table quietly decide a lot of outcomes. The same sitting rule means one weak sub-test invalidates three strong ones for GMC purposes. And grade B is not a single number everywhere — Australia now wants 360 in Speaking specifically, which is one ten-point increment above the bottom of the B band.

What actually happens in the OET Speaking sub-test

The shape is deliberately familiar. OET's own framing is that unlike tests which ask you to "respond to a computer, or talk about random topics like space exploration and recycling," the Speaking sub-test puts you in familiar territory: a clinical conversation.

Here is the sequence, start to finish:

• You are given a role-play card describing the setting, your role as the doctor, the patient's background, and the tasks you must cover.
• You get three minutes to read it, think, and make notes. You may write on the card and refer to it at any time while speaking.
• The interlocutor has their own card, with information you have not seen — including how the patient feels and what they will push back on.
• You speak for about five minutes. Then the whole thing repeats with a second, different card.
• The audio goes to assessors. Nothing is marked live, and the interlocutor is not your assessor.

Two 2026-specific details are worth knowing before you book. First, the sub-tests are profession-specific, so as a doctor you sit the Medicine version of both Speaking and Writing while Listening and Reading are shared across all twelve professions OET covers. Second, OET moved computer-based Speaking to a new delivery platform: from 9 June 2026, all new or rescheduled OET@Home and OET on Computer bookings run on a new secure OET Speaking Platform via Prometric, with a downloadable secure browser and a simplified ID check. OET states there is "no change to the format or marking" — only the plumbing changed.

If you want the card format itself pulled apart line by line, our companion piece on OET Speaking role play cards does exactly that, and the opening seconds get their own treatment in OET Speaking warm-up questions.

How OET Speaking is scored: the two criteria sets

This is where most candidates lose marks they did not know existed. OET marks your five minutes twice, against two completely different rubrics.

Scores from both sets are combined and converted to the 0–500 scale, per OET's results and scoring page.

And OET publishes the target. In the FAQ on its Speaking preparation page, OET states that "test-takers securing grade B will have achieved predominantly scores of 5 out of 6 on each linguistic criteria and 2 out of 3 for the clinical communication criteria." That is an unusually concrete benchmark for an exam body to print, and it reframes your preparation: you are not chasing perfection, you are chasing a consistent 5 and 2.

What does a 5 look like? OET's official Speaking assessment criteria and level descriptors define band 5 intelligibility as "easily understood," where "communication is not impeded by a few pronunciation or prosodic errors and/or noticeable L1 accent" and there is "minimal strain for the listener." Band 6 says an L1 accent "has no effect on intelligibility."

Read that twice, because it settles the question every international doctor asks. OET does not ask you to lose your accent. It measures whether your accent creates strain for the listener. A clearly-spoken Indian, Nigerian, Egyptian or Filipino accent scores band 5 all day. A rushed, mumbled imitation of an English accent does not.

The Clinical Communication side is even more explicit. The same document lists named indicators — initiating the interaction appropriately, adopting a non-judgmental approach, eliciting the patient's ideas, concerns and expectations, picking up the patient's cues, signposting changes in topic, using open questions before closed ones, not using compound or leading questions, establishing what the patient already knows before you explain, and checking they understood. Each is scored 3 for "adept use," 2 for "competent use," 1 for "partially effective use," and 0 for "ineffective use."

That list is essentially a Calgary-Cambridge consultation in rubric form. If you already consult well in your first language, most of the marks are yours — the risk is that English drops you from adept to partially effective for reasons that have nothing to do with medicine. We break the whole rubric down further in OET Speaking criteria and scoring.

The 0–500 scale and the grade bands

OET's Understanding your score guide publishes the band table:

Grade B is a range, not a point. A 350 and a 440 are both "B" on your certificate — but if you are heading to Australia, the difference between 350 and 360 in Speaking is the difference between registered and not.

What grade do doctors actually need for GMC and AHPRA

Three details in the GMC rules routinely catch people out, and all three are on the GMC's own using your OET certificate page:

• Same sitting, most recent sitting. You cannot mix a strong Speaking from March with a strong Writing from June. And a later, weaker attempt overwrites an earlier, stronger one.
• Two-year validity, measured against the decision date. For PLAB, the certificate must be valid on the day you sit PLAB part 1. For registration, it must be valid when the GMC approves your application — not when you submit it.
• There is an extension route, and it is narrow. If your certificate is over two years old and you passed PLAB 2 recently, the GMC may extend it — but only if the certificate is under three years old on the date your registration application was received and you passed PLAB 2 less than three months before that date. The sponsorship route gets no extension at all.

The Australian picture changed this year. The Ahpra accepted-tests page, reviewed 23 April 2026, states that National Boards updated minimum scores "to align with current score concordance research and scores set by the Department of Home Affairs for migration purposes." OET is now expressed numerically, and note the delivery restriction: on paper or computer based in a testing centre. If you sat OET@Home, check before you rely on it.

Anything you read on a recruitment blog about a different number is either out of date or about a different regulator. Open the regulator's own page, and check the review date at the bottom.

SpeakShark is a speaking-improvement tool, not an exam-prep provider, and is not affiliated with OET, Cambridge Boxhill Language Assessment, Pearson, the GMC, Ahpra, or any regulator or employer named on this page.

6 original OET doctors speaking role play scenarios

Every scenario below is written from scratch for speaking rehearsal. They are not OET role-play cards, they do not follow OET's card layout, and they contain no exam material. They are also not clinical guidance — follow your own protocols. What they do reproduce is the conversational pressure the criteria measure: a patient with a fixed expectation, an emotion you have to name, and information you have to deliver in lay terms.

Practise each as the doctor, out loud, against a partner or an AI that answers back.

1. The patient who wants a tablet, not a lifestyle conversation

A 52-year-old with a new type 2 diabetes diagnosis expects a prescription and a five-minute appointment. Your task is to explain the diagnosis and get to a shared plan that includes diet and activity.

• "Before I explain the results, can I ask — what were you expecting we'd find today?"
• "Your blood sugar has been running high for some time. What that means is your body isn't handling sugar the way it used to."
• "There is medication, and we may well use it. But it works far better alongside two or three changes at home. Can I take you through both?"
• "What part of that feels hardest to fit into your week?"

What the assessors hear: whether you established what the patient already knew before explaining (indicator E1), and whether you used open questions before closed ones. Front-loading a lecture costs you on information giving even when the medicine is perfect.

2. Declining antibiotics without losing the patient

A parent wants antibiotics for a child's sore throat that looks viral. They are tired, and they have taken a day off work to be here.

• "I can hear you've had a rough few nights, and you've taken time off to be here — thank you for bringing her in."
• "I've examined her carefully, and what I'm finding points to a virus rather than a bacterial infection."
• "Antibiotics won't shorten this, and they can give her side effects. What will help is regular fluids, and paracetamol at the right dose for her weight."
• "If she's not improving in three days, or if any of these things happen, I want to see her again — let me write them down for you."

What the assessors hear: non-judgmental attitude and relationship building under disagreement. The trap is a defensive register — "we don't give antibiotics for viruses" — which reads as a rule being quoted rather than a decision being explained.

3. An unexpected finding on a scan

A patient came in for back pain and the imaging has shown something that needs urgent further investigation. You do not have a diagnosis yet, and you must not invent one.

• "I've got the scan results. Before I go through them, is there someone you'd like with you?"
• "I'm afraid the news isn't as straightforward as we hoped." (pause — let three full seconds pass)
• "The scan has picked up an area we need to look at more closely. I can't tell you today what it is, and I don't want to guess."
• "What I can tell you is exactly what happens next, and when. Would that help?"

What the assessors hear: the warning shot, and your tolerance for silence. In a second language the urge to fill a pause with reassurance is overwhelming, and it destroys the pacing the fluency criterion rewards. This scenario gets a full treatment in our guide to OET Speaking and breaking bad news.

4. Consent for a procedure, in lay terms

You are consenting a nervous patient for an endoscopy. Every technical word you use is a word you then have to unpack.

• "The camera test looks at the lining of your stomach — it's a thin flexible tube with a light on the end."
• "You'll be awake but sleepy. Most people remember very little of it afterwards."
• "There are risks, and I want to be straight with you about them rather than have you find them on a leaflet later."
• "Can you tell me back, in your own words, what you'll be having done? I want to be sure I've explained it clearly."

What the assessors hear: appropriateness of language. Saying "oesophagogastroduodenoscopy" to a frightened patient is a linguistic error in OET's frame, not a display of knowledge — the criterion is explicitly about using words a non-medical person understands.

5. "Nobody has told us anything"

A patient's adult son arrives on the ward, angry, saying no one has explained his mother's situation. He is angry at you because you are there.

• "You're right to want answers, and I'm sorry you've had to chase them. Let me tell you what I know."
• "I can hear how worried you are about her."
• "Here's where we are today, and here's what we're waiting on."
• "Would it help if I arranged a time tomorrow when you, your mother and I can sit down together?"

What the assessors hear: picking up cues, and a steady low pitch. Under verbal pressure, second-language speakers speed up and rise in pitch, which reads as panic and costs you on fluency. Rehearse until your delivery stays flat and warm. If confrontation in English still spikes your heart rate, start with overcoming the fear of speaking English.

6. The telephone referral to a senior colleague

Not every OET role play is doctor-to-patient — but phone English is the weakest link for most international doctors regardless, and it is worth drilling on its own.

• "Hello, it's Dr Rahman, the medical SHO. I'm calling about a 68-year-old man on Ward 6 — I'm concerned he's septic."
• "Background: admitted two days ago with pneumonia, now spiking temperatures and hypotensive."
• "My assessment is he's deteriorating and needs review now."
• "Just to read that back to you — two litres of fluid, blood cultures, and you'll come within the hour. Is that right?"

What the assessors hear: providing structure — sequencing the interview purposefully and logically. On the phone your voice carries everything, which is why we gave it a separate guide: how to talk on the phone in English.

Run one scenario a day and rotate. On the free tier that is a focused five-minute rep at zero cost. Rehearse your first scenario free →

How to use your three preparation minutes

OET publishes its own checklist for the preparation window, and it is more useful than most paid advice. In its post on maximising your three preparation minutes, OET tells candidates to read the card for the patient's emotions first — looking for adjectives in the background information — then to anticipate what the patient expects the conversation to be about, then to identify where their expectation differs from your task.

That third step is the whole test in miniature. The gap between what the patient wants and what your card tells you to do is where the marks live. A rough three-minute routine:

• 0:00–0:45 — Read the setting and your task. Underline the two or three things you must cover.
• 0:45–1:30 — Find the emotion. Anxious? Frustrated? Sceptical? Write one word at the top of the card.
• 1:30–2:15 — Name the gap. What does the patient want that you are not going to give them, or the reverse?
• 2:15–3:00 — Plan your first two sentences only. Not a script — an opening. Everything after that is a conversation.

Do not write full sentences. Candidates who script themselves sound scripted, and hesitation while hunting for the next written line is exactly what the fluency descriptors punish.

Pronunciation traps that cost doctors intelligibility marks

Intelligibility is one of four Linguistic criteria and it is scored out of six, so a handful of habitual sounds can cap your Speaking grade regardless of how well you consult. Medical vocabulary is unusually cruel here — the spelling actively lies about the stress.

That last row is the one that matters most clinically and the one candidates most often ignore. Number stress is a patient-safety issue before it is an exam issue.

Do not try to fix all nine. Find the two or three that you actually produce wrong, and drill them inside full sentences rather than in isolation — the criterion scores pronunciation in connected speech, with prosody, not as a list of words. Real-time phoneme-level feedback is precisely the tool for finding which two, and nurses working through the same list will recognise the pattern in our English speaking practice for nurses guide.

How we researched this guide

Every number and rule above comes from a page we opened in July 2026, and each one is linked so you can verify it yourself.

Primary sources — the exam body. OET's Speaking preparation page for the twenty-minute length, the two five-minute role plays, and the grade-B benchmark of "5 out of 6 on each linguistic criteria and 2 out of 3 for the clinical communication criteria"; the Speaking criteria overview for the three-minute preparation window, the interlocutor's role, and the "at least two trained assessors" marking model; the official Speaking assessment criteria and level descriptors PDF for the band descriptors and the full indicator list; Understanding your score for the grade bands; results and scoring for the 0–500 scale and criteria counts; and the new Speaking Platform announcement for the 9 June 2026 migration to Prometric delivery.

Primary sources — the regulators. The GMC's own chapters on using your OET certificate and using your IELTS certificate; and Ahpra's accepted English language tests page, reviewed 23 April 2026.

A conflict we are showing rather than hiding. OET's own pages disagree about result turnaround. The test overview page says results are "typically available within 6 days for the OET Test on Computer, and OET@Home, and 12 days for the OET Test on Paper." The results and scoring page says computer and OET@Home results "can be made available within 48 hours of completing all four sub-tests," while a footnote on the same page says OET "aims to release results within 6 calendar days" and achieves it "in 95% of cases." Plan around the slowest figure, not the fastest, and never book a registration deadline against 48 hours.

What we deliberately did not do. We did not reproduce any OET role-play card, sample question or exam text. The six scenarios above are original scripts written for this article. We also did not print an Irish, Canadian or US requirement, because we could not open an authoritative regulator page for those during this research session — check your own regulator directly.

When OET Speaking practice is the wrong thing to do

An honest guide has to mark where its own advice stops.

You may not need OET at all. The GMC's guide lists several other routes to evidencing English: your primary medical qualification, a reference from an employer, a UK job offer, or other evidence. Before you pay for a test, read those chapters. Doctors spend money on OET every year that they did not have to spend.

If your gap is the exam, not the English, go to the exam body. SpeakShark is a speaking trainer. It does not teach the OET card format, it does not mark against OET's rubric, and it is not affiliated with OET. For format, criteria and genuine sample material — including full medicine role-play videos — use OET's own free resources and its two Ultimate Guides.

If one stubborn sound is your only problem, use a drill app. Tools built for isolated phoneme repetition with mouth-position graphics do a narrower job better than any conversational trainer, ours included.

If you are a nurse, not a doctor, the criteria are the same but the scenarios are not. Speaking is profession-specific; take the nursing route through OET Speaking for nurses instead.

If a different test suits you better, take it. Both the GMC and Australia's National Boards accept IELTS Academic; Australia also accepts PTE Academic, TOEFL iBT and Cambridge. If computer-scored speaking suits your nerves more than a live interlocutor, our PTE Academic speaking guide and the troubleshooting piece on why your PTE speaking score is low are the places to start.

And if you have not measured your baseline, do that first. A candidate at B1 does not need role-play tactics, they need months of daily conversation. Our free English speaking level test gives you an honest starting point before you book a test date.

How to build a daily OET Speaking habit

Reading criteria builds recognition. The exam measures production. Here is how the two connect on SpeakShark:

1. Open a Role Play session and set the scene. Role Play mode has the AI teacher play the patient — anxious, sceptical, angry — and respond to what you actually say. Because it reacts, you practise recovering when a consultation goes sideways, which no script can teach. More on why unscripted beats scripted in our role-play scenarios guide.
2. Let the phoneme feedback find your two sounds. Mispronounce malignant or fifteen mid-sentence and the exact sound is flagged in context, in real time — not as a vague note afterwards.
3. Let the memory do the coaching. This is the part no competitor in this niche has: SpeakShark remembers you between sessions — your recurring mistakes, your level, the scenarios you avoid. Tuesday's failed teach-back shapes Wednesday's session, instead of every app greeting you as a stranger and making you restart.
4. Match the accent to your destination. The free tier includes teacher Sarah (American English), three sessions of five minutes a day, 64 topics, no card. Premium — $10/month or $69/year, about $5.75/month — unlocks unlimited sessions, ten-minute conversations with unlimited turns, 320 topics across 10 categories, advanced feedback, full history and analytics, and all four teachers: American, British, Australian, Canadian. Heading for the GMC and the NHS? Rehearse with the British teacher.
5. Rotate weekly. Odd days: a high-pressure scenario (bad news, angry relative, phone referral). Even days: an explanation scenario (consent, diagnosis, medication change). Sunday: re-run whichever one went worst, which the app already knows.

Fifteen minutes a day beats a three-hour cram the weekend before. Start your free daily speaking session →

FAQ

The assessors are not listening for a doctor who has memorised phrases. They are listening for one who can hold a difficult conversation in English at conversational pace, name an emotion, explain a scan in plain words, and check they were understood — under time pressure, with a stranger. That is a reflex, and reflexes are built by speaking every day between now and your test date.

Start your free daily speaking session on SpeakShark → — three sessions a day, no card, with pronunciation feedback while you speak and a teacher that remembers what you got wrong yesterday.

All OET formats, criteria and score bands verified July 2026 from the official OET pages and PDFs linked above; GMC and Ahpra requirements verified the same week from the regulators' own pages. Requirements change — confirm with your regulator before booking. The six scenarios on this page are original practice scripts for communication rehearsal, not OET materials and not clinical guidance. SpeakShark is a speaking-improvement tool, not an exam-prep provider, and is not affiliated with OET, Cambridge Boxhill Language Assessment, Pearson, the GMC, Ahpra, or any regulator or employer named here.