Oxford and Cambridge Engineering Interviews
What the interview actually asks, and what a full answer looks like
What the interview actually asks, and what a full answer looks like
An Oxbridge Engineering interview is not a test of how much you know. It is a test of how you think. Tutors are not looking for polished answers to questions you have rehearsed — they are watching how you respond when you encounter something genuinely unfamiliar, how you apply physical intuition to a problem you have never seen, and whether you can be guided towards an answer through dialogue. If you are expecting something like a school oral exam or a university open day Q&A, you will be caught off guard. The questions are deliberately open-ended, often mathematical, and designed to push you beyond your comfort zone within minutes. That is not a flaw in the process — it is the point. Six of those questions are set out below, and one of them is then worked end to end in the three-part layout the General Engineering question pack uses.
Spring-and-pulley statics, a bulb-brightness circuit, and a charged particle curving through a magnetic field: an Oxbridge Engineering interviewer can open with any of these and expects reasoning from first principles, not a recalled formula. The General Engineering pack works through ten questions of that kind, and a free sample shows the format before you buy the rest of the pack.
The General Engineering pack — £180
Ten questions across 24 pages, each one carried through three stages: the bare question, then the hints to lean on if you stall, then a full worked answer. One PDF, one payment, instant download.
Get the General Engineering pack — £180Interviews for 2027 entry fall in December 2026 — Cambridge between the 7th and the 18th, Oxford across the month and online over Microsoft Teams.
Both Oxford and Cambridge Engineering interviews are conducted by academics — typically two interviewers — and last between 20 and 40 minutes. You may be shown a diagram, a physical object, or a short problem and asked to work through it in real time. The conversation is interactive: interviewers will prompt, redirect, and occasionally introduce new constraints to see how you adapt.
Oxford Engineering interviews tend to be highly mathematical and physics-focused, reflecting the analytical rigour of the course. Expect questions rooted in mechanics, calculus, and physical reasoning. Cambridge interviews, particularly for the Engineering Tripos, often place slightly more emphasis on breadth — you may encounter questions that blend physical intuition with design thinking or estimation. That said, the difference is one of emphasis rather than kind. In both cases, the core skill being assessed is the same: structured, confident reasoning under uncertainty.
Your ESAT score is in the room before you are
Interviewers at both universities see the test result before you sit down, and it shapes where the first problem starts. That is the practical reason the mechanics and the estimation you drill for the ESAT keep reappearing in the interview: the topics are the same and only the format changes — multiple choice against the clock on the test, spoken working in the room.
Both universities now shortlist Engineering applicants on the ESAT. The section immediately below sets out what the test covers and what it means for the year you are applying in.
You will almost certainly be asked something you cannot immediately answer. This is intentional. Tutors want to see whether you freeze or whether you engage — whether you can break a hard problem into smaller parts, make reasonable assumptions, and think out loud in a productive way.
Your admissions test result shapes how interviewers approach your session. Both Oxford and Cambridge Engineering applicants now sit the Engineering and Science Admissions Test (ESAT). Oxford replaced the PAT with the ESAT from 2026/2027 entry. The ESAT covers mathematics and physics through multiple-choice questions and is sat before interview. A strong ESAT score signals mathematical fluency and may lead interviewers to probe more deeply into analytical reasoning.
Preparing for the ESAT and preparing for the interview are not two distinct tasks. The problem-solving habits you build — working methodically, checking dimensions, drawing diagrams, applying fundamental principles — are exactly the habits that will serve you in the interview room. Tutors at both universities will have your ESAT result and may use it to shape which areas of physics and mathematics they explore with you on the day.
The most effective preparation combines three things: deepening your conceptual understanding, practising spoken problem-solving, and building the habit of intellectual curiosity.
On conceptual understanding: go beyond your A-level syllabus in the areas that matter most — Newtonian mechanics, energy methods, differential equations, and circuit analysis. You do not need to have studied university content, but you should be able to apply what you know flexibly and precisely.
On spoken problem-solving: this is where most candidates underinvest. Thinking aloud is a skill, and it feels unnatural at first. Practise working through problems verbally, narrating your assumptions, your method, and your uncertainty. Say things like "I'm going to assume the rod is uniform because otherwise I'd need more information" or "this doesn't feel right dimensionally — let me check." Interviewers find this far more compelling than silence followed by a correct answer.
Super-curricular preparation also matters. Reading engineering case studies, following developments in structural mechanics, aerospace, or materials science, and engaging with resources like the Isaac Physics platform or the Feynman Lectures will give you both intellectual depth and genuine things to discuss. Tutors notice when a candidate's curiosity is real.
A focused preparation plan might include:
Nothing on that list has to be bought: ESAT past papers are published, the reading is your own choice, and the mechanics revision is your A-level notes read with a harder eye. The part that is genuinely hard to arrange for yourself is the second one — working a problem you have never seen and then finding out, in writing, whether you set it up the way an interviewer would have wanted rather than merely whether the arithmetic came out. A model answer written at length does part of that job: it cannot watch you work, but it can show you the setup you should have reached and the check that would have caught you.
The following questions are representative of the style and difficulty you should expect. They are not trick questions — but they require more than recall.
The six questions above are the raw material. What a pack adds is the working, laid out in three parts and read in sequence: Questions, then Hints — which in Engineering usually means the interviewer changing one number and asking you to run the argument again — then Suggested answers, one continuous piece of reasoning in the first person, not a back-of-the-textbook solution that opens from the equation you were supposed to pick. The last of the six, the narrowing pipe, is worked below in that shape. One block has been added that is not one of the three parts: a first attempt that fails, so the prompts have something to catch.
Water flows steadily through a horizontal pipe that narrows from an internal diameter of 40 mm to 20 mm. Upstream, the water is moving at 1.0 m/s. What happens to the pressure at the narrow section, and roughly by how much? Would your answer change if the constriction were tighter still?
This is the answer the question is designed to draw out, and it is wrong in a useful way. “The pipe is narrower, so the water is being squeezed into a smaller space. Squeezing something raises its pressure — that is what happens when you put your thumb over a hose — so the pressure in the narrow section must be higher.”
The intuition is real, but the analogy is doing the wrong job. A thumb over a hose raises the pressure behind the thumb, not in the gap. Inside the gap the water is moving faster than it was, and something had to accelerate it. Once the acceleration is noticed the rest follows: the only thing that can push water along a horizontal pipe is a pressure difference, higher behind than in front. Answering “higher” skips the acceleration altogether, and that is the single observation the question is built on.
Start with continuity, because nothing else can be settled until the speed is known. The same volume of water crosses every cross-section each second, so A₁v₁ = A₂v₂. Area goes as the square of the diameter, and the diameter halves from 40 mm to 20 mm, so the area falls to a quarter and the speed has to rise by a factor of four: v₂ = 4 × 1.0 = 4.0 m/s.
Now the pressure. The pipe is horizontal, so no height term survives, and for a first pass I will treat the water as incompressible and ignore friction. Bernoulli then gives p₁ + ½ρv₁² = p₂ + ½ρv₂², so the drop is Δp = ½ρ(v₂² − v₁²) = ½ × 1000 × (4.0² − 1.0²) = 500 × 15 = 7500 Pa. The pressure at the narrow section is lower by about 7.5 kPa, not higher.
Is that a lot? Atmospheric pressure is around 100 kPa, so this is a fall of about seven and a half per cent — easily read on a gauge, nothing violent. Saying so out loud matters: the sense of scale is what makes the last part of the question worth asking.
Now take the throat down to 10 mm. The diameter is a quarter of the upstream value, the area a sixteenth, and the speed sixteen times as large: 16 m/s. The drop becomes ½ × 1000 × (16² − 1²) = 500 × 255 = 127,500 Pa, about 128 kPa. That is more than one atmosphere. If the upstream pressure is anywhere near atmospheric, the equation is now asking me to accept a negative absolute pressure in the throat, and no such thing exists — so the model has to break down before the flow ever gets there.
What breaks is the assumption that the water stays liquid. As the pressure in the throat falls towards the vapour pressure, the water boils at ordinary temperature, vapour cavities form, and they collapse again as the pressure recovers downstream. That is cavitation, and it is why pump inlets and propeller blades are sized with this calculation in hand. The honest way to close is to name the assumption that broke rather than the number that looked odd: continuity holds, because the mass still has to go somewhere, and Bernoulli holds as a statement about energy along a streamline. What does not survive is treating the fluid as a liquid the whole way through.
Not Bernoulli; nearly every candidate can quote it. What the question separates is whether you reach for continuity first — the step that turns a question about geometry into a question about speed — and whether you notice, without being told, that your own equation has stopped describing anything physical once the numbers change. The tighter constriction is asked for precisely because a model that predicts the impossible is the most productive thing that can happen in a twenty-minute conversation: it hands you something to defend and revise.
Which Engineering pack, and what is in it
Four sets face Engineering, ten questions each, printed the way the pipe question above is. Three of them are shelved under Engineering — General, Applied and Mathematical Engineering — and they work the same four areas at different tilts: mechanics and statics, thermodynamics and energy, circuits, and fluids. Take General first if you are only taking one. The fourth will not be where you go looking for it: the hub shelves Dynamics and Fields under Physics, and it is the set that reaches into electromagnetism, waves and optics.
£180 each, and the checkout takes 10% off two packs and 20% off three or more without a code. Engineering is one of the nine subjects that has a free sample file, so the format costs nothing to check: a question, the hints, and the solution.
Open the General Engineering pack — £180Reading about interview technique is useful, but nothing prepares you as effectively as working through a full exchange — a question, a first attempt, and then a stronger, more complete answer. The three exchanges below are representative of what candidates are actually asked in Oxford and Cambridge Engineering interviews for the 2025-26 and 2026-27 application cycles, covering the three areas that come up most often: order-of-magnitude estimation, structural reasoning, and basic circuit analysis.
Question: "Estimate the terminal velocity of a raindrop falling through air. You don't need to look anything up — work from what you know."
Weak answer: "I think it's about... maybe 10 metres per second? Raindrops fall pretty fast." This answer gives a number but no method. The interviewer has nothing to probe, no assumption to test, and no way to judge whether the candidate understands why a raindrop reaches a constant speed at all.
Strong answer: "A raindrop reaches terminal velocity when the drag force balances gravity, so I need to estimate both. Gravity is straightforward: weight equals density of water times volume times g — for a drop of radius 1mm, that's about 1000 kg/m³ × (4/3)π(0.001)³ × 9.8, which comes out to roughly 4×10⁻⁵ N. For drag, I'll use the standard drag equation, F = ½ρACd v², where ρ is air density (about 1.2 kg/m³), A is the cross-sectional area (πr² ≈ 3×10⁻⁶ m²), and Cd is roughly 0.5 for a sphere at this scale. Setting drag equal to weight and solving for v gives me v² = 2mg/(ρACd), and plugging in the numbers gives v somewhere around 7-9 m/s. That's consistent with what I vaguely recall about raindrops falling around 9 m/s, which is reassuring. If I made the drop bigger, the weight would scale with r³ while drag scales with r², so bigger drops should fall faster — which matches the fact that heavy rain has larger, faster-falling drops than drizzle." This answer works from first principles, states every assumption explicitly, sanity-checks the result against a known figure, and extends the reasoning to a related case — exactly the structure interviewers are trained to reward.
Question: "A simply supported beam of length L carries a single point load W at its midpoint. Sketch the bending moment diagram. Now tell me what happens to the maximum bending moment if you slide that load towards one of the supports."
Weak answer: "The bending moment is biggest in the middle, and I think if you move the load closer to a support it gets bigger there because that support has to carry more weight." This mixes up shear force and bending moment, and the conclusion — that moving the load towards a support increases the maximum moment — is the opposite of what actually happens, stated with confidence rather than tested.
Strong answer: "First I'll find the reactions. For a load W at distance a from the left support and b from the right (a + b = L), taking moments about the right support gives the left reaction as R1 = Wb/L, and by symmetry R2 = Wa/L. The bending moment at the point of the load is the reaction times the distance to it, so M = R1 × a = Wab/L. The diagram is two straight lines from zero at each support up to this peak at the load. Now, if I move the load towards, say, the left support, a gets smaller and b gets larger, but a + b is still L. The product ab is maximised when a = b = L/2 — that's basic algebra, since ab = a(L−a) is a downward parabola in a, peaking at the midpoint. So moving the load towards either support actually decreases the maximum bending moment, even though that support now carries a larger share of the reaction force. The reaction force and the bending moment aren't the same thing — the moment depends on the product of the reaction and the distance over which it acts, and shortening that lever arm outweighs the extra force." This answer sets up the statics properly, derives the general expression rather than quoting it, and correctly identifies — with a clear justification — that the intuitive answer is wrong.
Question: "You have a 9V battery and need to supply 3V to a small sensor. How would you design a circuit to do that, and what would you need to think about when choosing your components?"
Weak answer: "You'd use two resistors as a potential divider, with the ratio set so the output is a third of the input — so one resistor twice the size of the other." This identifies the right topology but stops at the textbook formula, with no discussion of what happens once a real sensor is connected.
Strong answer: "A potential divider is the right starting point: two resistors R1 and R2 in series across the 9V supply, taking the output across R2, gives Vout = 9 × R2/(R1+R2). For 3V out I need R2/(R1+R2) = 1/3, so R1 = 2×R2 — for example R1 = 20kΩ and R2 = 10kΩ. But the values I choose matter for two reasons the simple formula hides. First, power dissipation: with these values the current through the divider is 9V/30kΩ = 0.3mA, dissipating under 3mW, which is fine — but if I'd chosen much smaller resistors to make the output more stable against loading, I'd be wasting significant power as heat, which matters if this is battery-powered. Second, and more important: the sensor itself has some input resistance, and it sits in parallel with R2 once connected. If the sensor's resistance is comparable to or smaller than R2, it will pull the output voltage down below my calculated 3V — this is the 'loading effect'. Avoiding it means choosing R1 and R2 much smaller than the sensor's input impedance, which pushes me back towards the power-dissipation problem, so there's a genuine trade-off to manage. If the sensor draws meaningful current or needs a precisely regulated 3V regardless of load, a linear voltage regulator IC would be a better choice than a simple divider, because it actively holds the output voltage constant rather than depending on the ratio of a fixed pair of resistors." This answer gives the textbook circuit, then goes further to identify the two things that make the naive version fail in practice — exactly the applied reasoning Engineering interviewers are testing for.
The most damaging mistake is silence. Candidates who go quiet when they do not know the answer give interviewers nothing to work with. Even a partially formed thought, spoken aloud, is more useful than a polished answer that never arrives. If you are stuck, say so — and then say what you do know that might be relevant.
A second common error is over-rehearsing answers to anticipated questions. Interviewers can tell when a candidate is reciting rather than reasoning, and they will quickly move to territory you have not prepared. Depth of understanding always outperforms breadth of memorised content.
Candidates also frequently neglect units and physical sense-checking. If your answer gives a bridge a mass of 400 kilograms or a cyclist a power output of 50 kilowatts, you should notice. Dimensional analysis and order-of-magnitude reasoning are not optional extras — they are core engineering habits that tutors actively look for.
Finally, do not treat the interview as adversarial. The interviewers are not trying to humiliate you — they are trying to find out how you think. Engage with them as collaborators in a problem, not as examiners to be satisfied.
Twenty to forty minutes, and a partial answer beats a silent one every time
The worked exchanges above all turn on the same moment: a first attempt that was not right, said out loud, which gave the interviewer something to build on. Candidates who wait for certainty never reach that moment.
The pipe question earlier on this page is built that way round: the wrong answer first, then the two sentences that would have moved you off it, then the working, with the failed model named rather than quietly dropped. Every question in the Engineering sets is printed to that order, at £180 each.
See the four Engineering packsMost Engineering interviews at both Oxford and Cambridge last between 20 and 40 minutes. Oxford candidates often have two separate interviews, sometimes with different tutors or colleges, so the total time in interview can be longer. Cambridge candidates may also have more than one interview depending on the college. Each session is focused and moves quickly — there is rarely time for extended discussion of any single topic.
Not exactly — but you may be taken beyond what you have covered. Interviewers will often introduce a new constraint or extension to a problem to see how you reason when the familiar framework runs out. You are not expected to know university-level content, but you are expected to apply A-level concepts in unfamiliar ways. The skill being tested is reasoning, not recall.
The single most effective method is spoken mock interviews with someone who can give you honest, specific feedback — not just on whether your answers are correct, but on how clearly you communicate your reasoning. Practising problems silently on paper does not replicate the conditions of the interview. You need to build the habit of thinking aloud under mild pressure, and that only comes from repeated practice in a realistic setting.
Say so — and then engage anyway. Tell the interviewer what you do know that might be relevant, what approach you would try, and what is making the problem difficult for you. Interviewers are experienced at guiding candidates who are genuinely trying, and a candidate who engages openly with uncertainty is far more impressive than one who bluffs or falls silent. The interview is a conversation, and your willingness to think through difficulty is itself part of what is being assessed.
A Fermi estimation question asks you to derive an approximate value for a physical quantity — like the terminal velocity of a raindrop or the power output of a cyclist — using only reasoning, known constants, and sensible assumptions, without looking anything up. Interviewers use them because they strip away rehearsed knowledge and expose how you actually think: whether you can identify the right physical principle, state your assumptions explicitly, work through the algebra cleanly, and sanity-check your final number against something you already know. A confident guess with no method scores far lower than a modest number reached through clear, stated reasoning.
Start from statics, not intuition. Draw a free-body diagram, find the support reactions using equilibrium, then derive the bending moment as a function of position rather than trying to recall a formula. Many candidates get structural questions wrong not because they don't know the theory, but because they trust an intuitive guess — for example, assuming a larger reaction force always means a larger bending moment — over the algebra. Working through the derivation aloud, step by step, lets the interviewer see exactly where your reasoning is strong and gently correct you if a step goes wrong, which is the point of the exercise.
Beyond the correct formula, interviewers want to see that you understand why a simple circuit can behave differently once it's connected to something real. A potential divider calculation is only the first step — the strongest candidates go on to discuss loading effects, power dissipation, and the trade-offs involved in choosing component values, and can suggest a better alternative (such as a voltage regulator) when the simple approach has real limitations. This applied, practical layer of reasoning — anticipating how theory meets a real device — is exactly what distinguishes a strong Engineering interview answer from a merely correct one.
Yes, and the position for candidates preparing now is: Cambridge Engineering applicants have already been sitting the ESAT since 2025 entry, replacing the NSAA and ENGAA, and this continues for 2026 entry. Oxford is completing its own transition — the PAT's final sitting was for 2026 entry, and Oxford Engineering Science applicants now sit the ESAT from 2027 entry onwards, with registration opening 20 July 2026 and the test itself sitting 12–16 October 2026. In practice, this means any candidate applying for 2027 entry to either university's Engineering course should be preparing for the ESAT, not the PAT.
Ten Oxbridge Engineering interview questions, each set out in three parts read in sequence: Questions; then Hints, which in Engineering usually means the interviewer changing one number and asking you to run the argument again; then Suggested answers, one continuous piece of reasoning in the first person rather than a back-of-the-textbook solution. Four sets face Engineering: General, Applied and Mathematical Engineering are shelved under Engineering on the resources hub, and Dynamics and Fields under Physics, so it is easy to miss. They are £180 each. The pipe question above is printed in that same shape, so the format is not something you have to take on trust.
General Engineering is ten Oxbridge Engineering interview questions, each one written out as the question, the interviewer’s prompts, and the answer in full. £180, a PDF you download and keep.
Open the General Engineering pack — £180Written by specialist subject tutors. 91% of our students achieve their desired grades. Rated Excellent on Trustpilot. That is the company’s rating, not a rating of the packs.
No. Ten questions spread over four areas — mechanics and statics, thermodynamics and energy, circuits, and fluids — is two and a half per area, which is not syllabus coverage and was never sold as it. For sheer volume the published ESAT material is free and should be your first stop, though be warned that the archive is short: the test is only a couple of years old, so there is nothing like the stack of past papers an older exam would give you. What a pack holds that a question bank does not is the argument written out at length underneath each question — the setup, the follow-ups you would have been given, and the check at the end. Use the published material for repetition, and the pack for the part no mark scheme records.
Yes. Every pack is £180, and the reduction is applied in the checkout rather than by code: 10% off two packs, 20% off three or more. Taking General, Applied and Mathematical Engineering together therefore clears the 20% threshold on its own. The basket is not organised by subject, so anything else you add from the hub counts towards the same threshold.
Partly, and not in the obvious way. The ESAT is multiple choice against the clock; a pack is written answers with no time limit, so it is not test drill. What carries across is the setup: naming your assumptions before you lean on them, keeping the algebra symbolic long enough to see what cancels, and asking whether the answer is the right size. Those are the moves that decide an ESAT question as much as an interview one. For timed practice, use the published past papers, and see the ESAT preparation page for what the test itself involves.