American admissions advice talks about "the Oxbridge interview" as if Oxford and Cambridge run the same process under two names. They don't. Both are real, both are more structured than the "weird trick question" reputation suggests, and both are genuinely different mechanisms -- verified here against Oxford's own department admissions data and Cambridge's own published statistics and admissions pages, not third-party test-prep summaries. If you're looking for how to actually prepare -- including Oxford's TARA and Cambridge's TMUA admissions tests -- see our interview and admissions-test prep guide alongside this breakdown of the process itself.

Two different systems, not one

Oxford shortlists before it interviews. Every applicant sits a subject admissions test (which one depends on the course), and shortlisting happens after that, before anyone is interviewed. Usually 2 to 3 applicants are shortlisted for every place available. Being shortlisted is what gets you an interview at all.

Cambridge interviews first and sorts afterward. Cambridge interviews the large majority of applicants directly -- its own admissions pages describe interviewing "all applicants with a realistic chance" of an offer -- and only after interviews does it run the Winter Pool, redistributing strong applicants who impressed a college without space for them.

That order matters. An Oxford applicant who isn't shortlisted never gets an interview at all. A Cambridge applicant almost always gets interviewed, and the sorting that decides which college actually admits them happens afterward, in January, largely out of sight.

Oxford: shortlisted, then interviewed twice

Oxford's own Computer Science admissions feedback for the 2025-26 cycle -- a real, department-published document, not an estimate -- shows exactly how this works in practice: 531 applicants, 144 shortlisted (27%), 58 offered a place (11% of all applicants, 40% of those shortlisted).

The part most American applicants don't expect: every shortlisted Oxford applicant is interviewed by two colleges, not just the one they applied to. Each applicant keeps their first college (chosen, reallocated, or randomly assigned for open applications) and is separately assigned a second college by an algorithm designed to balance interview load across colleges. In the 2025-26 Computer Science cycle, most applicants were interviewed twice by their first college and at least once more by their second.

There's also a pre-interview safety valve working in the opposite direction: if one college ends up with an unusually high number of shortlisted applicants per place, some of those applicants are randomly reallocated to a different college before interviews even happen. In December 2025, 25 Computer Science applicants were reallocated this way -- and 10 of them went on to receive an offer from the college they were reallocated to. Interviews themselves were almost entirely remote by video call for this course and cycle, often using a shared virtual whiteboard for problem-solving.

Decisions aren't locked to whichever college interviewed you, either: 28.2% of Computer Science offers that cycle were either open offers or made by a college other than the applicant's first college -- part of a deliberate university-wide effort to admit the strongest applicants regardless of which college happened to interview them.

Cambridge: interviewed first, possibly pooled after

Cambridge's own Undergraduate Admissions Statistics for the 2025 cycle give the university-wide real numbers: 22,513 applications, 4,893 offers (21.7%), 3,669 acceptances -- a 16.3% overall success rate. For Computer Science specifically: 1,739 applications, 163 offers (9.4%), 128 acceptances (7.4%).

Most Cambridge applicants have 1 or 2 interviews totaling 35 minutes to an hour, with 2 or 3 interviewers -- sometimes including an academic from outside the applicant's own subject. Where those interviews happen isn't uniform: it depends entirely on which college is assessing you, from fully online to fully in-person to a mix that depends on where you live.

After interviews, Cambridge runs its Winter Pool: about 18% of applicants from the October 2025 cycle were pooled -- made available to other colleges with remaining places, because their own college was impressed but didn't have room. Pooling isn't a rejection -- in January 2026, 1,060 pooled applicants, roughly 1 in 4, received an offer from a different college than the one they applied to.

<!-- chart:oxbridge-interview-funnel -->

What interviewers actually say they're testing for

Both universities push back directly on the "impossible trick question" reputation. Oxford's own guidance states plainly that interviews aren't about tricks or mind games, and that tutors aren't judging how fast an applicant reaches a "right" answer -- they're watching how an applicant responds to new ideas and unfamiliar material, the same way a subject tutorial works once you're actually enrolled. Cambridge's own list of what it assesses is similarly concrete: understanding of the subject, readiness for a high academic level, the ability to think critically and independently, and genuine curiosity -- not a specialist body of trivia.

Oxford scores this on a real, structured basis, too: each interview is judged against agreed descriptors on a numeric scale, shared among every tutor involved in that applicant's admission, alongside their UCAS form and test results -- not a single interviewer's private impression.

What this means for you

  • Know which system you're actually walking into. Not being shortlisted at Oxford ends your application before interview; not receiving an Oxford or Cambridge offer after a Cambridge interview can still mean a Winter Pool offer is coming in January.
  • At Oxford, expect two colleges' worth of interviews, not one. Preparing for "the interview" as a single event undersells what shortlisted applicants actually go through.
  • At Cambridge, check your specific college's interview format in advance -- some conduct every interview online, some entirely in person, several split by where the applicant lives.
  • Neither university is testing for a "correct" fast answer. Both say directly that they're assessing how you think through something unfamiliar, scored against real criteria, not testing recall or timing.

Sources