Skip to content

OET Speaking: The Complete Guide for Nurses

The format, all 9 assessment criteria with real examples, a full worked roleplay, common mistakes, and how to prepare — in one place.

By SpeakOET Team · July 29, 2026

If you’ve sat OET Speaking before and walked out thinking “I talked the whole time, why wasn’t that enough?” — you’ve already found the thing that trips up most candidates. Speaking isn’t judging whether you can hold a conversation in English. It’s judging whether you can run a clinical conversation — structured, reactive, patient-centred — while also speaking clearly. Most candidates prepare for the language half and never train the other half at all.

Time
~20 minutes (2 roleplays, ~5 min each + prep)
Marks
9 criteria, converts to Grade A–E, 0–500 scale
Parts
2 roleplays with a trained interlocutor
Difficulty
High — live, unscripted, no second take
Passing grade
Usually Grade B / 350 — confirm with your regulator
Who this is for
Nurses nervous about live roleplay, or repeat candidates

The fastest way to understand OET Speaking is to do one.

Try a Free Roleplay Now

What is OET Speaking?

OET Speaking tests the exact conversation you have dozens of times a shift — explaining a procedure to a nervous patient, gathering a history from someone in pain, reassuring a relative without over-promising. It isn’t a general English speaking test with a medical topic bolted on; it’s a simulation of your actual job, recorded and scored against how real clinical communication is supposed to work.

That’s why fluent, confident English speakers still sometimes score lower than expected. Talking smoothly for five minutes isn’t the target — structuring the conversation, picking up on what the patient actually says, and checking they’ve understood are all scored separately from how clearly you pronounce your words.

Why confident speakers still lose marks

A nurse who talks fluently but delivers a memorised, one-directional explanation without pausing to check the patient’s understanding can score well on the linguistic criteria and still lose marks across most of the clinical communication criteria — which make up 60% of your final band.

It also plays out very differently from a general English speaking test. IELTS Speaking is an interview: an examiner asks you questions about your life, your opinions, a familiar topic, and you answer as yourself. OET Speaking is a roleplay: you stay in character as the nurse for the full five minutes, the “examiner” is acting as a patient who might get upset, go off-topic, or push back, and you’re expected to manage that the way you would on a real ward. Candidates who prepare as if it’s an interview tend to answer questions well but struggle to drive the conversation themselves.

The format

OET Speaking is taken separately from the other three sub-tests, usually on a different day. You sit with a trained interlocutor — in person or over video for OET@Home — and complete two roleplays, each about 5 minutes long. Before each one you get roughly 2–3 minutes to read a card describing the scenario: who you are, who the patient or relative is, and the points you need to cover.

The interlocutor plays the patient and improvises around the scenario rather than following a script, so the conversation genuinely depends on what you say. Both roleplays are recorded and sent to two independent examiners, who each score you against the same 9 criteria. Your final grade is the average of the two.

StageWhat happens
Prep time~2–3 minutes to read the card and plan, per roleplay.
Roleplay 1~5 minutes with the interlocutor, recorded.
Roleplay 2~5 minutes, a different scenario, recorded.
AssessmentTwo examiners independently score both recordings later.

All 9 assessment criteria

Examiners score you against two groups of criteria, each 0–6. Knowing them by name changes how you practise — instead of vaguely “sounding fluent,” you can target the exact thing losing you marks.

Clinical communication criteria (60% of your band)

1. Relationship building

Greeting the patient, using their name, and acknowledging how they feel before diving into facts or questions.

Weak Facts first

“Right, Mr. Patel, I need to ask you about your chest pain.” Correct information, zero acknowledgement of the person you’re talking to.

Good Person first

“Hello Mr. Patel, I understand you’ve been having some chest pain — that must be worrying. I’d like to ask a few questions so we can look after you properly.”

2. Understanding and incorporating the patient’s perspective

Picking up on what the patient actually says — a worry, an unexpected detail — rather than delivering your own points regardless of their response.

Weak Ignoring the cue

Patient: “I’m scared it’s the same thing that happened to my father.” Nurse: “Okay, so next I’ll explain the procedure.” The fear is left completely unaddressed.

Good Responding to the cue

“I can understand why that would be frightening, given what happened with your father. Let’s talk through what we know about your situation specifically.”

3. Providing structure

Signposting what’s coming next, so the patient (and the examiner) can follow where the conversation is going.

Weak No signposting

Jumping straight from questions into advice with no transition, leaving the patient unsure whether the conversation has moved from gathering information to a decision.

Good Clear signposting

“I’d like to ask a few questions first, then explain what happens next.” One sentence, and the whole roleplay becomes easier to follow for everyone.

4. Information gathering

Open questions first to let the patient talk, closed questions to confirm specifics, then summarising back what you’ve heard.

Weak Closed questions only

“Is the pain sharp? Is it worse at night? Have you taken paracetamol?” — efficient, but the patient never gets to say anything you didn’t already predict.

Good Open, then closed

“Can you tell me more about the pain?” followed by targeted follow-ups once the patient has described it in their own words.

5. Information giving

Explaining a diagnosis, procedure, or advice in a way the patient can actually follow — as suggestions, not orders — and checking they’ve understood.

Weak Jargon, no check

“You’ll need a course of prophylactic antibiotics post-operatively.” Technically correct, and the patient has no real idea what that means or whether they agree.

Good Plain language, checked

“We’ll give you some antibiotics after the surgery to help prevent infection — does that make sense, or would you like me to go over it again?”

Linguistic criteria (40% of your band)

6. Intelligibility

Pronunciation, word stress, and rhythm clear enough that a listener doesn’t have to work to understand you — not the same as having zero accent.

How to improve

Record yourself explaining a common procedure and listen back specifically for words you rush or swallow — usually a handful of recurring words, not your speech in general.

7. Fluency

Speaking at a natural pace with natural pauses, without long silences or a stream of “um” and “ah” while you search for words.

How to improve

A short thinking pause is completely normal and doesn’t cost marks — it’s the filler-word habit that fills every pause that examiners notice, since it obscures whether you’re actually forming a clear thought.

8. Appropriateness of language

Register suited to a patient conversation — professional but plain, with medical terms explained rather than assumed.

Weak Wrong register

“We’ll need to cannulate you and start you on IV fluids for your hypovolaemia.” Correct clinically, unintelligible to most patients.

Good Right register

“We’re going to put a small tube into a vein in your arm so we can give you fluids directly, since you’re a bit dehydrated.”

9. Resources of grammar and expression

Range and accuracy of grammar and vocabulary under real-time pressure — varied sentence structure, not just correctness.

How to improve

Under pressure, most candidates default to short, simple sentences on repeat. Practising out loud — not just reading — is what builds the reflex to vary structure without stopping to think about grammar mid-sentence.

The clinical communication half is what trips up strong English speakers most often — fluent grammar doesn't score well if you talk over the patient or forget to check they've understood.

Practise against all 9 criteria

SpeakOET's AI patient improvises like a real interlocutor and scores every roleplay against the same 9 criteria above.

Try a Free Roleplay

A step-by-step exam strategy

Speaking doesn’t run on a clock you manage minute by minute the way Reading does — but each stage still has a clear job, and most lost marks come from skipping one.

StageWhat to do
Reading the cardIdentify who you are, who the patient is, what they’re likely worried about, and the 2–4 points you must cover. Note key words, not full sentences.
Planning your orderDecide roughly which point comes first, second, third — the card rarely lists them in the ideal spoken order.
OpeningGreet the patient by name and acknowledge their situation before asking or explaining anything.
Middle of the roleplayWork through your points, but react to whatever the interlocutor actually says — an unexpected question is not a distraction from the task, it is part of it.
ClosingCheck understanding and confirm next steps before the interlocutor ends the roleplay — don’t let it just trail off.

Decision tree: interlocutor says something unexpected?

Does it directly relate to one of your card’s required points?

Yes → address it as part of that point, then continue your planned order.

No → briefly acknowledge it (don’t ignore the patient), then bridge back: “that’s a good question, I’ll come back to it — first let me just check…”

If you go blank mid-roleplay

A short pause to think costs you nothing. What costs you marks is abandoning the conversation’s structure entirely — take the pause, then return to your next planned point rather than improvising something unrelated out of panic.

A full worked example

Here’s a realistic (invented) roleplay card, followed by two short excerpts of how the same moment can go — one losing marks, one scoring well — with the reasoning for each.

Roleplay card

Setting: Orthopaedic ward, day 1 after hip replacement surgery.

Patient: Mrs. Chen, anxious about her first physiotherapy session this afternoon.

Your tasks:

— Explain why early mobilisation after hip replacement is recommended

— Address her fear of falling or the new hip “giving way”

— Explain briefly what will happen during the physio session

— Check she feels ready and confirm next steps

Weak Weak opening — task-first, card followed like a checklist

“Mrs. Chen, this afternoon you have physiotherapy. It’s important to mobilise early after hip surgery to prevent complications like blood clots and stiffness. The physiotherapist will help you stand and take a few steps.” All three facts are correct. Nothing has acknowledged that she’s frightened, and the tasks were delivered as one uninterrupted block rather than a conversation.

Good Strong opening — person first, then structure

“Hello Mrs. Chen, how are you feeling this morning? I know you have physiotherapy this afternoon, and I wanted to talk that through with you, since I understand you might be feeling a little nervous about it. Is that fair to say?” The patient’s likely emotional state is named before any clinical content, and the sentence ends with a genuine question rather than moving straight on.

Why the strong version scores higher: it hasn’t sacrificed any information — both versions eventually cover the same three tasks. The difference is entirely in relationship building and providing structure, two of the five clinical communication criteria, which is exactly why two candidates who “covered everything on the card” can still score very differently.

The same pattern repeats through the rest of the roleplay: when Mrs. Chen says “I’m scared it might just give way while I’m walking,” a card-following response moves straight to explaining the physio session; a strong response answers the fear directly first (“that’s a very common worry, but the new joint is designed to bear your weight from day one, and the physiotherapist will be right beside you”) before continuing. Same information, delivered as a response to the patient rather than a script running on schedule.

18 mistakes that quietly cap your score

None of these will make the roleplay collapse — they just quietly cap a score that otherwise sounds fluent and confident.

  1. Launching straight into facts without acknowledging the patient first. Costs relationship building before you’ve said anything clinically wrong.
  2. Delivering a memorised script instead of reacting to what the interlocutor says. Examiners can tell within seconds when a candidate is running a rehearsed monologue.
  3. Using clinical jargon without checking the patient understood. Correct terminology scores nothing if the patient is left confused.
  4. Missing one of the card’s required points under time pressure. Usually caused by poor prep-time planning, not running out of things to say.
  5. Talking without structure. No signposting means the examiner can’t follow where the conversation is going, even if every fact is correct.
  6. Asking only closed questions. Efficient, but it never lets the patient describe things in their own words — losing information-gathering marks.
  7. Ignoring an emotional cue. A patient naming a fear or worry that gets no acknowledgement at all before you move on.
  8. Giving advice as an order rather than a suggestion. “You need to do X” instead of “I’d recommend X, how does that sound?”
  9. Never checking understanding. Explaining something once and moving on, with no “does that make sense?” moment anywhere in the roleplay.
  10. Filling every pause with “um” or “ah.” A brief silent pause to think reads as natural; a constant verbal tic reads as disfluency.
  11. Rushing to finish within 5 minutes. Speeding through the card to make sure everything gets said, at the cost of intelligibility and structure.
  12. Treating an unexpected question as an interruption. Deflecting or ignoring it instead of briefly addressing it, which directly costs the patient-perspective criterion.
  13. Over-preparing one fixed opening line. A line that doesn’t fit the specific card still gets used, and it visibly doesn’t match the scenario.
  14. Speaking only in short, simple sentences throughout. Safe, but it caps the resources-of-grammar-and-expression criterion, which rewards range.
  15. Not using the patient’s name. A small detail that repeatedly signals relationship building across the whole conversation.
  16. Ending abruptly once the last card point is covered. No closing check or confirmation of next steps, so the conversation just stops.
  17. Practising only with a script, never live improvisation. Builds confidence with memorised lines that collapses the moment a real interlocutor deviates from them.
  18. Only practising roleplays you find easy. Avoiding scenarios involving bad news, anger, or confusion means the first time you face one for real is in the actual exam.

Tips from experienced OET tutors

  • Build a flexible opening “shape” (greet, name, acknowledge, signpost) rather than a fixed sentence — it adapts to any card instead of sounding rehearsed on the ones it doesn’t fit.
  • Record every practice roleplay and listen back once for content, once purely for filler words — the two passes catch different problems.
  • Practise scenarios you’d normally avoid: an angry relative, a patient refusing treatment, delivering unwelcome news. These come up, and the exam is the wrong place to try them for the first time.
  • During prep time, write down the patient’s likely emotional state in one word before anything else — it forces relationship building to happen early instead of being an afterthought.
  • Ask a practice partner (or an AI patient) to genuinely improvise and interrupt, rather than sitting silently while you deliver your card — the improvisation is the actual skill being tested.
  • Don’t memorise full answers to common scenarios. Memorise structure (open → gather → explain → check) and let the specific words come from the specific card.
  • Get feedback against the 9 named criteria specifically, not general impressions — “that sounded good” from a friend doesn’t tell you whether you lost marks on structure or on patient perspective.
  • Practise standing up or sitting the way you would in a real consultation, not slouched at a laptop reading notes off-screen — posture and delivery habits carry over into how you sound.
  • If a practice roleplay goes badly, redo the exact same card a day later rather than moving straight to a new one. Fixing the specific breakdown matters more than accumulating new scenarios.

Always check the exact current format, timing and marking guide on the official OET website — those details are set by OET, not by us.

Frequently Asked Questions

What is the OET Speaking sub-test?

OET Speaking is a face-to-face (or video-call) roleplay test taken separately from Listening, Reading, and Writing. You play yourself as a nurse; a trained interlocutor plays a patient or relative. It's profession-specific, so nurses get nursing scenarios, and it's recorded for two examiners to assess afterwards.

How is OET Speaking scored?

Two examiners independently score your recording against 9 criteria, split into linguistic criteria (intelligibility, fluency, appropriateness of language, resources of grammar and expression) and clinical communication criteria (relationship building, understanding the patient’s perspective, providing structure, information gathering, information giving). Scores convert to a 0–500 scale and a letter grade A–E.

How long is the OET Speaking test?

About 20 minutes total: two roleplays of roughly 5 minutes each, with about 2–3 minutes to read each card before you start. The exact timing is set by OET, so confirm current figures on the official site before your test date.

What’s a good OET Speaking score for nurses?

Most regulators, including the UK NMC, ask for at least Grade B (350/500). See our full breakdown of score requirements by regulator, or check your target country’s exact number with the score calculator.

Can I fail Speaking while passing the other three sub-tests?

Yes — each of the four sub-tests is graded independently, and most regulators set a minimum for every one of them. A strong Writing or Reading score does not compensate for a weak Speaking score.

Is the interlocutor a real patient or an actor?

A trained interlocutor, not a real patient — but they play the role convincingly and improvise around the scenario rather than reading from a script. Treat them exactly as you would a real, anxious patient or relative, because that improvisation is exactly what the clinical communication criteria are assessing.

Can I ask the interlocutor to repeat something?

Yes, and doing so naturally — 'sorry, could you say that again?' — costs you nothing. It's a completely normal thing to say to a real patient too, and examiners are listening for how you communicate, not for a flawless first-time hearing.

Do I get to choose which two roleplays I do?

No, the two scenarios are assigned to you as part of the test. They're drawn from realistic nursing situations, so broad familiarity with common ward, clinic, and community scenarios matters more than trying to predict the exact topic in advance.

What happens if I run out of things to say?

A short pause to think is completely normal and won't cost you marks on its own — real conversations have pauses. What does cost you is going silent for so long the interlocutor has to rescue the conversation, or panicking and abandoning the card's remaining points altogether.

Is OET Speaking harder than IELTS Speaking?

They test different skills. IELTS Speaking is an interview about general topics and opinions; OET Speaking is a roleplay where you perform your actual job — explaining, reassuring, gathering information — in character. Most nurses find OET Speaking more natural once they stop treating it like an English interview. See our full OET vs IELTS comparison for more.

How can I practise OET Speaking without a study partner?

Roleplay practice is the one part of OET prep that’s genuinely hard to do alone, since you need someone improvising as the patient. An AI patient that responds in real time and scores you against the 9 criteria solves that — that’s exactly what SpeakOET’s speaking practice does.

Is OET@Home Speaking different from the in-person test?

The format and criteria are identical; only the delivery changes (a live video call for OET@Home, face-to-face with an interlocutor for the in-person test). Always confirm current delivery options on the official OET website, since these are set by OET, not by us.

Ready to try a real roleplay?

Free AI patient roleplay with instant scoring against all 9 official criteria and detailed feedback on exactly what to fix.

Start Free Speaking Practice