The Common Data Set (CDS) is a standardized questionnaire that colleges complete each year, publishing their admission, enrollment, and financial-aid numbers in one consistent format. For applicants it is the primary source: acceptance rates, test-score ranges, and exactly which factors each college says it weighs.
Nearly every reliable admission statistic you have ever seen — on a rankings site, in a guidebook, or in our simulation — traces back to a Common Data Set filing. Learning to read the original takes about twenty minutes and permanently upgrades your information diet: no more secondhand numbers of unknown vintage, no more forum folklore. This page covers what the CDS is, the handful of sections worth your time, and how to find one.
The CDS runs from section A (general information) through J (degrees conferred), but almost everything an applicant needs lives in section C — first-time, first-year admission:
| Section | What it contains | Why you care |
|---|---|---|
| C1 | Applicants, admits, and enrollees, by gender | Divide admits by applicants for the real acceptance rate; enrollees by admits for the yield rate. |
| C7 | Importance grid for 19 admission factors | The college tells you, in writing, what it weighs — from essay to alumni relation to level of interest. |
| C9 | Test-score ranges and submission shares | The middle-50% SAT/ACT of enrolled students, plus what fraction actually submitted scores. |
| C10–C12 | Class rank and GPA distributions | Where admitted students really sat in their classes — the reality check for your transcript. |
| C21–C22 | Early Decision and Early Action counts | Where the real ED and EA acceptance rates come from — when the college fills them in. |
| H | Financial aid | Average need-based awards, the share of need met, and merit-aid counts. |
Two worked examples of what these sections yield. From Harvard's filing: 54,008 applicants, a 3.64% overall acceptance rate, a 1510–1580 SAT middle-50%, a 3.95 average unweighted GPA, and a C7 grid that rates eight factors "very important." From Amherst's: both alumni relation and level of applicant's interest are marked "not considered" — a legacy applicant and a five-time campus visitor get exactly nothing for it there, and the college put that on the record itself. Our guide to demonstrated interest maps that C7 factor across all 192 colleges we track.
Search the college's name plus "common data set." The filing usually lives on the institutional research or provost's office site as a PDF or spreadsheet, often with a decade of archives beside it. If search fails, try the college's own site search for "CDS," or go directly to its institutional research office page.
Three caveats for when you get there. Not every college publishes: some, especially public universities, skip years or omit sections — most publics leave the C22 Early Action counts blank, which is why EA rates are hard to pin down at state schools. Filings lag: the current admissions cycle's numbers will not appear until the following year's CDS, so check the cover year. And definitions matter: C9 describes enrolled students, not admitted ones, and under test-optional policies only the submitters among them — both details push published ranges higher than the full admit pool.
Three high-value moves, in order of effort:
College Monte Carlo uses the same filings as machine input: the per-college acceptance rates, early-round splits, test ranges, and C7 factor grids in our simulation are compiled from each college's Common Data Set and cross-checked against primary sources — the full inventory is on our data sources page, and the modeling assumptions on the methodology page.
Whenever you meet an admission statistic without a source, ask: what does the CDS say? If the number cannot be traced to section C of a college's own filing, treat it as rumor with confidence intervals unknown.