Turn the table around for a minute. Stop thinking about what you're going to say and think about what the people opposite you have been asked to do.
They have read your case study, probably on a train, probably alongside eleven others. They have your PEDRs, your CV and your exam scripts. And they have somewhere between twenty minutes and an hour to decide whether the person in front of them can be let loose on the public as a registered architect. That is the whole job. Not to humiliate you, not to test whether you memorised the Construction Act, but to check that the competent, decisive professional described in the written submission is the same person now sitting in the chair.
Everything that happens in that room follows from that. Once you understand the examiner's problem, the questions stop feeling random.
Format varies by provider, so check your own handbook rather than trusting anyone's war story — including this one. But the shape is broadly consistent: a panel of two or three examiners, usually a mix of practising architects and someone from the course, a table, your submission documents in front of them, and a conversation that moves through your case study and out into general practice.
One of our tutors sat his at Chester racecourse, in the directors' boxes, with the whole cohort lined up in a long hall waiting to be called. His description: the pre-waiting area is full of chatter about what everyone put for question four, then you go upstairs and the entire room goes silent, and you sit there watching a door, waiting to get dragged in. He was last. Alone with his thoughts for the best part of an hour.
That corridor is the worst part of the day, and it is the part you can do nothing about. Plan for it. Bring something to read that isn't your case study — rereading it at that point changes nothing and makes you sound rehearsed. Eat beforehand. Accept that you will be nervous and that nerves are not a scoring criterion.
The thing our tutors all say about the room itself is that it settles within the first two exchanges. You are nervous walking in, and then someone asks you a normal question about your project, you answer it, and you remember that you know this project better than anyone else alive. From there it's a conversation between professionals. That's not a motivational line — it's what the format is designed to produce.
Four Things, and Only Four
Strip out the specifics and a Part 3 oral is testing four things.
Contractual understanding. Not clause recital. Whether you know which contract you were working under, why that form was chosen over the alternatives, what it obliged you to do, and what it obliged everyone else to do. If you administered a contract, the panel wants to hear the mechanics in your own words: how a change was instructed, what the valuation process was, what happened when the programme slipped.
Professional judgement under pressure. The scenario questions. Client asks you to issue a practical completion certificate on a building that isn't finished. Contractor goes insolvent mid-contract. You spot a defect that your own drawing caused. These aren't trick questions. They're testing whether you reach for the right order of operations — duty, contract, insurer, client — rather than the most agreeable answer.
Honesty about what went wrong. Every project has a mess in it somewhere. A panel that can't find the mess assumes you're hiding it, and a candidate who presents a flawless project has either not understood the project or is lying. The critical analysis in your case study is supposed to be critical. The oral is where they check you meant it.
Whether your answers match your paperwork. This is the one candidates underestimate, and it's the hinge the whole thing turns on.
Your Submission Is the Prosecution's Evidence
The examiners have your PEDRs. Your PEDRs say, in your own handwriting, what you did and when, countersigned by your employer. Your case study says who made which decisions. Your CV says what your role was.
If you say in the room that you ran the pre-start meeting, and your PEDR for that quarter shows you were observing site activity rather than leading it, that is a problem. Not because you've committed a crime, but because the panel now has to decide which version is true — and they are being asked to certify you on the basis of documents you signed.
So before anything else: reread your own PEDRs. Properly. Most candidates write them in a rush at quarter-end, months apart, and have genuinely forgotten what they claimed. Line your submission up against your case study narrative and find every place where the two describe the same event differently. Fix the account in your head now, not under questioning.
And if there's a genuine discrepancy — your role grew, the record lagged behind, the wording was loose — say so first, before they find it. "My PEDR for that quarter understates it, because I was logging hours rather than responsibility" is a fine answer. Being caught out and then explaining is a much worse one.
The Second Question Is the Real One
Here is how a panel finds out whether you ran a project or watched one.
They ask something open. You answer. Then they ask a follow-up that only someone who was actually in the room could answer. Then another. The first question is a doorway; the second and third are the test.
"We issued an extension of time of four weeks." Fine. Who assessed it? What was the relevant event? What did the contractor claim and what did you award? Did you consult anyone? What did the client say when you told them? Did loss and expense follow, and how was that valued? Who signed the certificate?
That chain is the whole technique. People who led the work answer it fluently, including the bits that reflect badly on them, because they remember the argument. People who were adjacent to the work start generalising around question three. The tells are predictable: the pronoun slides from "I" to "we" to "the practice", the answers get shorter and more procedural, and specifics are replaced with what should happen rather than what did.
Run that chain on yourself. For every significant decision in your case study, you want to be able to answer six things cold: who decided, on whose advice, under which contractual mechanism, who carried the risk, what it cost in time or money, and what you'd do differently. If any of those six comes back blank, that's where the panel will land.
Three Ways to Lose It in Ten Minutes
Claiming a decision you didn't make. The most common unforced error. You were in the meeting where the director decided to recommend a change of procurement route, and by the time you've told it three times in rehearsal it has become your recommendation. Don't. A panel that catches one overclaim starts auditing everything else you've said, and you spend the rest of the session on the back foot.
Hedging on who held the risk. Risk allocation questions are deliberately uncomfortable, because risk allocation is uncomfortable. Who carried the design risk on the cladding package? Who was responsible for the contractor's design portion? Whose insurance would have responded? If you waffle here, the panel reads it as not understanding your own project's structure. The honest answer — including "we transferred it under the contract and in my view that was the wrong call" — always beats the diplomatic one.
Being vague on insurance and appointment terms. What was your practice's PI cover and what was the excess? What was in your appointment, what was excluded, what were the net contribution and liability-cap provisions? Was it a RIBA standard form, a bespoke client document, or an email? Candidates who have never read their own practice's appointment documents get exposed here fast, and it's the easiest gap in the world to close. Go and ask for the appointment. Read it. Ask your director why the cap is what it is.
The Project You Didn't Really Lead
Plenty of candidates come to Part 3 with a case study where their role was real but partial. That is normal, and panels know it. What they will not accept is the inflation.
The way through is to be precise about the boundary. "I didn't chair the design team meetings — the project architect did. I prepared the agenda, ran the drawings register, issued the information and chased the consultant responses, and from stage five I was administering the contract under her supervision." That tells the panel exactly what you can be examined on, and it sounds like someone who understands the difference between contributing and leading — which is, in itself, a professional judgement.
Then bring your judgement to the parts you did own. A small decision you genuinely made, with its consequences, is worth more than a large decision you watched.
Say You Don't Know, Properly
You will be asked something you can't answer. The project detail you've forgotten, the statutory point you never dealt with, the contract form you've never used.
There is a good version of this answer and a bad one. The bad one is guessing, confidently, and then compounding it when the panel pulls the thread. The good one has three parts: say what you don't know, say what you'd do about it in practice, and say what you do know that's adjacent. "I've not administered a design and build contract, so I'd be relying on the practice's standard procedures and on advice before I committed to anything. What I can tell you is how the equivalent worked under the traditional contract I did administer, and where I'd expect the differences to bite."
That is a competent architect's answer. Knowing the limits of your own competence and acting accordingly is a professional obligation, not a weakness — and a panel is far more comfortable registering someone who demonstrates it than someone who bluffs.
Your Own Project Is the Question Bank
Stop revising in the abstract. Build the rehearsal out of your own case study.
Take the project chronologically, stage by stage, and write down every decision point — appointment, procurement route, planning strategy, the moment the budget moved, the contract award, the first instruction, the delay, the defect, handover. For each one, answer the six questions above out loud. Not in your head. Out loud, to another person, because the gap between "I know this" and "I can say this" is where orals are lost.
Then get someone to ask you the second question. And the third. A study partner from your cohort is ideal — our tutors describe forming a three-person group after the crash course and running sessions several evenings a week for months before the exam, stepping the intensity up in the last four to six weeks. The value isn't the content. It's that someone else chooses where to drill.
Two more things worth doing. Find out who's on your panel, and look them up — one tutor researched his examiners in advance so he could pitch the conversation to their specialisms, which is sound practice, although printing their photographs and dropping them on the table mid-interview is, on reflection, not. And prepare three short project narratives you can deploy: one where you made a call and it worked, one where you made a call and it didn't, one where you escalated because it was beyond your competence. Thirty to sixty seconds each. Not scripts — spines, so you don't ramble.
On length generally: answer the question, stop, and let them steer. Overlong answers are the most reliable sign of a nervous candidate filling silence, and every extra sentence is another door for a follow-up you didn't want.
The Thing Worth Remembering
A referral at the oral is not a verdict on your intelligence. It's usually a verdict on a gap — in experience, in evidence, or in the match between the two. Providers handle referral and feedback differently, so read your own course's assessment criteria rather than inheriting someone else's version of it; the broader requirements and provider differences are covered in our guide on what to know before you apply.
What's in your control is this. Know your project to the depth of the third follow-up question. Know what your appointment and your insurance actually said. Be the same person on the page and in the chair. And when you don't know, say so like a professional rather than a candidate.
You ran the project. Go and talk about it like someone who did.
