A student can look up a school's overall acceptance rate, feel reasonably confident about their odds, and then get rejected from an engineering program at that same school -- not because their profile was weak, but because the number they checked wasn't actually the number that applied to them. At a growing list of schools, engineering (and computer science specifically) isn't admitted the same way as the rest of the university, and the gap between "the school's rate" and "the engineering program's rate" can be enormous.

Why: most selective engineering programs admit by college or major, not by university

At many schools, you don't apply to "the university" and declare a major later -- you apply directly to a specific college or school within it, and that choice is binding at the point of application, evaluated against a separate, usually much smaller and stronger applicant pool. This is the same structural mechanism covered in our piece on switching into a capacity-constrained major after enrolling -- the reason switching in later is hard at schools like Michigan is the same reason getting in as a freshman is harder in the first place: a fixed number of faculty, lab seats, and equipment that can't expand just because more students want in.

The numbers, school by school

UC Berkeley admits directly by college, and the gap is stark: the university's overall admit rate has run in roughly the 10-11% range in recent cycles, while Electrical Engineering and Computer Sciences (EECS) -- a joint program housed in the College of Engineering -- has admitted under 5% of its applicants.

Carnegie Mellon shows one of the widest internal gaps of any school that publishes broken-out data: its School of Computer Science has admitted under 5% of applicants, while Dietrich College of Humanities and Social Sciences, at the same university in the same cycle, has admitted roughly 24%. Applying "to Carnegie Mellon" isn't really one admissions process -- it's several, with very different odds depending on which college you apply to.

The University of Washington admits directly to major for computer science and computer engineering through the Paul G. Allen School, and the gap by residency has been reported as particularly wide: in the 2025 cycle, the direct-to-major admit rate ran around 37% for Washington residents but dropped to roughly 5% for nonresidents and about 2% for international applicants, against an overall university admit rate closer to the high-40s percent range -- see the full three-year breakdown, straight from UW's own published data for the trend and why there's no GPA target to aim for.

University of Illinois Urbana-Champaign publishes its own first-choice-major admit rate for every college on campus: Computer Science sits at 7.4%, against 21.2% for the Grainger College of Engineering overall -- CS is roughly a third as selective as its own parent college. See the full breakdown of UIUC's official admit-rate table for how every other college compares.

Georgia Tech doesn't publish a separate rate for its College of Engineering or College of Computing, but the university has been explicit that it admits by intended major and that those two colleges are among its most sought-after -- meaning its official overall rates (28% in-state, 9% out-of-state, for the Class of 2030) function as a ceiling, not a real estimate, for anyone applying to engineering or computer science specifically.

Worth being precise about what's confirmed here and what isn't: Berkeley, CMU, and UW figures above come from data those schools or campus outlets have reported specifically for engineering/CS programs; Georgia Tech's engineering-specific rate isn't officially published at all, so treat any specific percentage you see cited for it as an outside estimate, not a confirmed number.

This isn't unique to engineering

The same "advertised number isn't the number that applies to you" pattern shows up in pre-med, where a school's headline acceptance rate often describes only students who already cleared an internal committee screen, and in nursing, where national capacity -- faculty and clinical placement slots, not applicant quality -- is the actual bottleneck. Engineering's version of the same story is structural too: it's rarely that engineering applicants are held to a subjectively higher bar than other applicants at the same school, it's that a fixed number of seats in a specific college is being divided among a specific, usually stronger and larger, pool of people who all want that exact major.

It's also not always the hardest major at a given school -- Business and Public Health both admit at a lower rate than Computer Science at UC Berkeley specifically, and the gap between a competitive major and its campus's overall rate scales with how selective the campus already is, from nearly nonexistent at UC Merced to 36x at UC Irvine's Nursing major.

What this means for you

  • Look up the admit rate for the specific college or major you're applying to, not just the university's headline number, before you decide how confident to feel about a school -- the gap between the two can be several times larger than the headline rate itself.
  • If a school admits by college or major, that choice is often binding at the point of application. Don't apply "undecided" hoping to switch into engineering later without checking that specific school's policy first -- some, like Michigan, close that door entirely once you're enrolled.
  • A school with a high overall acceptance rate isn't automatically an easy admit for engineering. Purdue, UIUC, and Georgia Tech-style schools can have comfortable-looking headline rates and a genuinely competitive engineering-specific bar underneath them.
  • When a school doesn't publish an engineering-specific rate, treat any number you find quoted for it as an estimate, not a confirmed figure -- the same caution that applies to any admissions statistic a school hasn't released itself.

Sources