Elite Careers StrategyWho Gets In?A research programme published by Elite Careers Strategy
London
Evidence date 10 September 2026
Published 10 September 2026
Research lead: Hassan Akram
For journalistshassan@accessecs.comInterview or comment within three working daysData checks same dayPress kitDownload reports
How to quote this pageQuote the coverage before the ratio. Thirteen of thirty is the finding; the range is a property of those thirteen records.

Check a figure
Elite university selectivity

What thirty of the world's most highly ranked universities publish about admissions

Complete institution-wide undergraduate admissions figures were verified for 13 of the 30 universities in the QS World University Rankings 2027 top-30 band. The remaining cases reveal substantial differences in what global admissions systems count, publish and make comparable.

Sampling frame: QS World University Rankings 2027, published 18 June 2026. All sources retrieved and reverified 10 September 2026.

Census coverage

13complete verified pairs
11partial records
4structurally non-comparable
2not located by this review
Sampling frameEvery institution placed 1 to 30 in the QS World University Rankings 2027. Tied positions retained. QS defines which universities to look at and nothing else.

The record itself is the finding

Every year a family is told that a place at a particular university is close to impossible to win. Almost always the number attached to that claim is an admission rate. Almost never is anyone told what that rate counts.

Elite Careers Strategy set out to test whether a like-for-like admissions figure can be assembled across the world's most highly ranked universities. We took the thirty institutions placed from 1 to 30 in the QS World University Rankings 2027, published on 18 June 2026, and went to each institution's own published record. No QS score, indicator or rank position was used as evidence about admissions.

For thirteen we verified a complete institution-wide undergraduate pair: a published application count and a published count of positive admissions decisions, from the same institution, for the same cycle. For seventeen we did not. That is the result.

BoundaryFailure to verify a complete pair does not establish that a university withholds the data, or that the data do not exist.

Why seventeen records are not complete

Four reasons. The institution publishes part of the pair but not all of it. The official data we needed was not accessible to us. No pair could be located. Or the university operates an admissions architecture in which an institution-wide rate is not a meaningful object at all.

That last category deserves emphasis. At ETH Zurich and EPFL, a domestic applicant holding the right school-leaving qualification is entitled to enter, so there is no institution-wide competitive selection to measure. At the National University of Singapore and Nanyang Technological University, admission happens at programme level on national examination results. At Peking, Tsinghua and Fudan, places are allocated against province and programme quotas through the gaokao. An admission rate for these institutions would not be a smaller number than Harvard's. It would be a category error.

What ECS records is what one review, conducted on 10 September 2026 from publicly reachable official sources, was able to confirm.

Five architecturesUS institution-wide admissions, UK conditional offers, qualification-based entry, programme-level selection and centralised allocation cannot be treated as one process.

The spread is architecture before it is competition

Institutions count different things. Some count applicants; others count applications. Some publish offers; others publish admissions. Those are not the same object, and this page does not silently convert one into the other.

A single example shows why the decision point matters. Cambridge published 4,893 offers for the 2025 cycle and 3,669 acceptances. A UK offer is conditional on examination results still to come. A US admission is not conditional in the same way. Reading a UK offer rate against a US admission rate treats two different stages of two different processes as one measurement.

Every complete record located in this review came through one of three comparatively structured reporting environments: the US Common Data Set or official institutional reporting ecosystem, the UK's UCAS-linked reporting system, or McGill's voluntary publication of its admissions funnel. That is a finding about reporting infrastructure, not about candour.

01
The verified records

Thirteen complete records, grouped by admissions architecture

These figures are not a league table of difficulty. Institutions count different units, make decisions at different stages and operate under different admissions architectures.

Read with that warning in front of it: among the thirteen universities with complete records the reported positive-decision rate ranged from 3.80% to 47.55%, equivalent to approximately 2.1 to 26.3 applications or applicants per recorded positive decision. That range is a property of thirteen records drawn from five different admissions architectures. It is not an ordering, not a difficulty scale and not a ranking.

US institution-wide, applicants and admissions

One institution-wide application, counted as applicants. The decision is an unconditional admission. Figures come from each institution's Common Data Set Section C1 or its own published admissions statistics.

InstitutionCycleCounting unitApplications or applicantsPositive decisionsDecision unit, in the institution's own wordsApplications or applicants per recorded positive decision
Stanford University ECS-derivedFall 2025 (CDS 2025-2026)Applicants60,6462,302who were admitted (1,162 + 1,140 + 0)26.34
Harvard UniversityClass of 2029, autumn 2025Applicants47,8932,003Admitted (includes 75 admitted from the waiting list)23.91
Princeton UniversityFall 2025 (CDS 2025-26)who applied42,3031,868Total first-time, first-year (degree-seeking) who were admitted22.65
Johns Hopkins UniversityFall 2025 (CDS 2025-26)who applied50,2593,072Total first-time, first-year (degree-seeking) who were admitted16.36
Cornell UniversityFall 2025 (CDS 2025-2026)Applicants72,5236,077Total first-time, first-year students who were admitted (published total, C.117)11.93
University of California, BerkeleyFall 2025 (CDS 2025-26)who applied126,86414,524Total first-time, first-year (degree-seeking) who were admitted8.73

US institution-wide, applications and admissions

The same reporting ecosystem, but the denominator counts applications rather than applicants. An application count and an applicant count are not the same object and are not converted here. The decision each institution publishes differs too, and each row states it in that institution's own words.

InstitutionCycleCounting unitApplications or applicantsPositive decisionsDecision unit, in the institution's own wordsApplications or applicants per recorded positive decision
Yale UniversityFall 2025 (Year Entered 2025, Class of 2029)Total Applications50,2652,388Total Admits21.05
Massachusetts Institute of TechnologyClass of 2029, autumn 2025Applications29,2811,334Offers of admission21.95
University of PennsylvaniaFall 2025, Class of 2029Applications72,5443,570were offered admission20.32

UK UCAS, applications and conditional offers

A UK offer is conditional on examination results still to come, so it sits at a different stage from a US admission. Cambridge published 4,893 offers for the 2025 cycle and 3,669 acceptances.

InstitutionCycleCounting unitApplications or applicantsPositive decisionsDecision unit, in the institution's own wordsApplications or applicants per recorded positive decision
University of Oxford2025 cycle (UCAS applications by 15 October 2024, mostly for entry October 2025)Applications23,3293,891offers made6.00
University of Cambridge2025 cycle (applications from September 2024)Applications22,5134,893Offers4.60

UK UCAS, applications and admissions

Imperial publishes admissions rather than offers, so its ratio is applications to admissions. It is not an offer rate and is not directly comparable with Oxford or Cambridge.

InstitutionCycleCounting unitApplications or applicantsPositive decisionsDecision unit, in the institution's own wordsApplications or applicants per recorded positive decision
Imperial College London2025-26 entryApplications32,2143,259New admissions 2025-26 entry9.88

Other voluntarily published institutional funnels

McGill publishes an institution-wide applicants, offers and entering class funnel by choice rather than under a shared reporting standard. Its figure is an offer rate, not an admission rate, its offers may include conditional offers, and its undergraduate count bundles professional-faculty entry.

InstitutionCycleCounting unitApplications or applicantsPositive decisionsDecision unit, in the institution's own wordsApplications or applicants per recorded positive decision
McGill UniversityFall 2025Applicants38,13518,132Offers2.10
One record is marked ECS-derived. Stanford's Common Data Set prints no total line for Section C1, so its totals are an ECS calculation from official component rows. Stanford publishes males, females and students of unknown sex, so those three rows are exhaustive and their sum is the population total. Every component, the addition and the reconciliation check are published in the evidence briefing, and the derived total is not the sole support for any finding on this page. Yale's Common Data Set has the same shape but prints only men and women, so a derived total would have been a floor; Yale's figures here are instead the directly published totals from its Office of Institutional Research report W033, which are one applicant and one admit higher than the Common Data Set rows sum to.
Source Each institution's own published admissions record, retrieved and reverified 10 September 2026. Full URLs, locators, access dates and arithmetic are in the evidence briefing.
Not publishedNo global selectivity league table, ordering or difficulty scale.

What this evidence cannot support

No defensible global admissions-selectivity league table can be constructed from these records. Seventeen institutions lack a verified complete pair, and the thirteen complete records span materially different units, decision points and admissions architectures.

QS rank and published undergraduate admissions ratios measure different things. The admissions figures here must not be read as a proxy for ranking position or institutional quality. No association between rank and admissions ratio is calculated, because thirteen records spanning five admissions architectures and two counting units cannot support one.

We are publishing the gaps rather than filling them with estimates.

02
The next gate

What ECS could not verify on employer recruitment pages

ScopeThirteen named employers. One review date. One evidence standard. Quoting this finding without that scope misstates it.

Across the public recruitment pages of 13 named elite employers reviewed by ECS on 10 September 2026, ECS could not verify both an application-volume figure and a programme-intake figure covering the same programme, geography and cycle.

The thirteen employers reviewed were JPMorgan Chase, Goldman Sachs, Morgan Stanley, Bank of America, Citigroup, McKinsey & Company, Bain & Company, Boston Consulting Group, Freshfields, Clifford Chance, A&O Shearman, Linklaters and Slaughter and May.

This does not establish that the figures do not exist, that the employers deliberately withhold them, or that their programmes are more selective than the universities reviewed.

No numerical comparison between any university and any employer is made here, because no employer numerator and denominator on equivalent terms was available to compare.

CommentaryAttributed interpretation. Not a dataset finding.

Commentary

For families, entry to an elite university is one selection gate, not the end of selection. The next system is often less transparent.

Hassan Akram, Founder and Principal Advisor, Elite Careers Strategy

The first sentence is interpretation and carries no evidential claim. The second is supported only within the named employer-review scope set out above and in the methodology.

EvidenceEvery URL, locator, access date and arithmetic step is published in the briefing.

Method, evidence and citation

Title Who Gets In? Elite University Selectivity
Corporate author Elite Careers Strategy
Research lead Hassan Akram, Founder and Principal Advisor
Published 10 September 2026
Revised 10 September 2026
Version 1.0
Sampling frame QS Quacquarelli Symonds (2026) QS World University Rankings 2027. Available at: topuniversities.com/world-university-rankings (Accessed: 10 September 2026). QS defines the sampling frame only.
Methodology /who-gets-in/methodology/
Evidence briefing Who Gets In? Elite University Selectivity, evidence briefing, version 1.0 (PDF)
Source register /who-gets-in/register/

Suggested citation

Elite Careers Strategy (2026) Who Gets In? Elite University Selectivity. Research lead: H. Akram. London: Elite Careers Strategy. Available at: https://www.accessecs.com/who-gets-in/selectivity/ (Accessed: [insert the date on which you accessed this page]).

Machine assistance. Source retrieval, arithmetic, reconciliation checks and drafting for this page were carried out with AI assistance under human direction. Every figure is traceable to a named official source with a URL, locator and access date, and every calculation is published. Editorial responsibility rests with Elite Careers Strategy.

Commercial interest. Elite Careers Strategy is a paid advisory firm working in the field this research examines. The research uses public institutional sources only and contains no client evidence, but the commercial interest is disclosed rather than left to be inferred.