Colleges report the share of their class sections with fewer than 20 students. Across 233 colleges with 2025-26 Common Data Sets, the median reported share is about half. The median estimated share of course enrollments in those sections is about 23 percent, and the median ceiling, under the most generous assumptions, is about 35 percent.
When a college says that 70 percent of its classes have fewer than 20 students, what share of its students’ course enrollments are in those classes?
Pomona College and Harvard University make the question concrete. In their 2025-26 Common Data Sets, Pomona reports 74.3 percent of its fall 2025 class sections with fewer than 20 students and Harvard reports 73.6 percent, nearly the same figure. But Harvard also ran 49 sections with 100 or more students, while Pomona ran none. Weighting each section by the number of students it holds, College Transitions estimates that 55.3 percent of Pomona’s course enrollments were in sections under 20, compared with 28.2 percent of Harvard’s.
The published figure counts sections. Every class counts once, whether it holds 12 students or 300, and the large sections hold far more of the enrollments. To measure how much that matters, we compiled the class-size tables from the 2025-26 Common Data Sets of 233 colleges, covering nearly 395,000 undergraduate sections taught in fall 2025, and estimated how course enrollments were distributed across them.
Across the 233 colleges, the median share of sections with fewer than 20 students is 50.1 percent. The median estimated share of course enrollments in those sections is 22.9 percent. The estimate rests on assumptions about how many students sit in each section, so we also calculated a mathematical ceiling using only the reported size bands and section counts. It is the highest share those counts allow. Across the 233 colleges, the median of those ceilings is 34.7 percent.
What the 20-Student Figure Counts
The number comes from item I-3 of the Common Data Set, a standardized report that many four-year colleges publish each year. Each college sorts its undergraduate class sections for the fall term into seven size bands, from 2 to 9 students up to 100 or more, and reports how many sections fall in each band. Rankings and admissions materials often cite one statistic drawn from it, the percentage of sections with fewer than 20 students.
Three parts of the CDS definition shape what that percentage can tell a family.
The unit is the section. A school with 900 sections of 15 students and 100 sections of 200 students reports 90 percent of its classes under 20, even though the 100 large sections hold 20,000 enrollments and the 900 small ones hold 13,500.
Labs, recitations and discussion sections are reported in a separate subsection table and left out of the headline figure. A 300-person lecture that meets weekly in 20-person discussion groups counts as one section of 100 or more; the discussion groups appear only in the subsection table. Those small meetings are part of many students’ weeks and are not reflected in the share of sections under 20.
The top band has no ceiling. A section of 100 students and a section of 600 land in the same column.
Admissions materials use the same kind of section count, sometimes with a different cutoff. UCLA’s admissions site says that “nearly two out of three undergraduate classes have 30 or fewer students,” and its 2026 viewbook says “two out of three undergraduate classes have fewer than 30 students.” UCLA’s 2025-26 CDS supports that kind of claim, with 62.9 percent of its fall 2025 sections in the bands below 30. By our estimate, those sections held 20.6 percent of UCLA’s course enrollments, and the ceiling is 32.5 percent. Because the CDS groups classes of 30 to 39 students together, we can verify the viewbook’s “fewer than 30” claim but not the admissions site’s “30 or fewer” figure precisely. None of these figures is directly comparable with the under-20 numbers elsewhere in this article.
Counting Enrollments Instead of Sections
Each section in a size band holds some number of students between the band’s limits. Multiplying the number of sections in each band by the band’s midpoint (14.5 for the 10-to-19 band, for example) estimates the course enrollments in that band. Sections of 100 or more have no midpoint, so the main estimate counts 150 students in each. Dividing enrollments in the two smallest bands by all enrollments gives the estimated share of course enrollments in sections under 20.
These are course enrollments, not students. A student taking four courses counts four times, once in each section. The estimates describe how enrollments are spread across the sections a college reports. They do not describe any individual student’s schedule, which depends on major, year and course choices. The CDS also counts any section with at least one degree-seeking undergraduate, so some mixed-level courses include graduate students.
Figure 1 plots the estimate against the reported section share for all 233 colleges.
Figure 1. Reported share of fall 2025 undergraduate class sections with fewer than 20 students, and the estimated share of course enrollments in those sections, 233 colleges. Source: Common Data Set 2025-26, item I-3, compiled by College Transitions; enrollment shares are College Transitions estimates.
Every college sits below the dashed line, which marks equal shares of sections and enrollments. That much is arithmetic. A small section holds fewer students than a large one, so small sections always account for a smaller share of enrollments than of sections. The chart shows how large the drop is. At 139 of the 233 colleges, the estimated enrollment share is less than half the reported section share.
The two measures still rank colleges in a similar order (r = 0.92), so the reported figure remains useful for comparing schools. As a guide to where course enrollments sit, though, it overstates the small-section share at every school in the group.
Testing the Assumptions
The overall median changes relatively little under other assumptions about the largest sections. Counting every section of 100 or more at exactly 100 students, the lowest value possible, moves the median estimate from 22.9 percent to 23.9 percent. Counting them at 250 moves it to 21.7 percent.
Individual colleges with many large lectures are more sensitive, and Table 1 shows how much. It reports the estimated share of enrollments in sections under 20 under three assumptions, along with each college’s ceiling, which places every small section at the top of its band (9 or 19 students) and every larger section at the bottom of its band (20, 30, 40, 50 or 100). The ceiling needs no estimate of how large the 100-plus sections are, because it counts each at the band’s minimum of 100. The last column shows the estimated share of enrollments in sections of 100 or more under the main assumption of 150 students.
| College | Under 20, 100+ at 100 | Under 20, 100+ at 150 | Under 20, 100+ at 250 | Ceiling | In 100+ sections (at 150) |
| Pomona College | 55.3% | 55.3% | 55.3% | 66.7% | 0.0% |
| Harvard University | 31.3% | 28.2% | 23.6% | 44.0% | 29.4% |
| Stanford University | 30.4% | 27.7% | 23.5% | 42.5% | 26.8% |
| University of Pennsylvania | 21.5% | 19.7% | 16.9% | 32.0% | 24.8% |
| University of Michigan | 19.4% | 17.3% | 14.1% | 28.2% | 33.3% |
| University of California, Los Angeles | 12.7% | 10.7% | 8.1% | 19.2% | 47.1% |
| University of Texas at Austin | 14.2% | 12.8% | 10.7% | 21.5% | 29.6% |
| Texas A&M University | 7.7% | 6.9% | 5.7% | 12.1% | 31.5% |
Table 1. Sensitivity of the estimates for eight colleges. Source: Common Data Set 2025-26, item I-3; College Transitions estimates.
The gap between Harvard and Pomona holds under each assumption tested. Even if each of Harvard’s 100-plus sections held exactly 100 students, its estimated small-section share of enrollments would be 31.3 percent, against 55.3 percent at Pomona, whose estimate does not change because it reported no sections of 100 or more. UCLA’s figures for large sections vary the most. Depending on the assumption, sections of 100 or more hold between 37 and 60 percent of its estimated enrollments. A college with many lectures of 300 or more students would place even more of its enrollments in that band than the 250-student column shows.
A note on the 34.7 percent figure: it is the median of the 233 individual ceilings, not the ceiling of the college with the median estimate.
A Few Large Sections Hold Many of the Enrollments
Two kinds of totals appear in this article. Medians describe the college in the middle of the group, so each college counts once. Pooled figures add every section across all 233 colleges, so large universities, which report thousands of sections, carry more weight. Because large universities tend to report lower shares of small sections, the pooled share of sections under 20 (43.6 percent) is lower than the median share (50.1 percent).
Pooled across all colleges, sections under 20 held an estimated 15.7 percent of course enrollments. Sections of 100 or more were 4.9 percent of all sections and held an estimated 23.0 percent of enrollments. Looking at each college separately, the median share of sections with 50 or more students is 8.4 percent, and the median estimated share of enrollments in those sections is 30.8 percent.
Figure 2 breaks the estimated enrollments at 14 colleges into five class-size groups, with Pomona and Harvard at the top.
Figure 2. Estimated distribution of course enrollments by class size, fall 2025, 14 selected colleges, with each college’s reported share of sections under 20 at right. Source: Common Data Set 2025-26, item I-3, compiled by College Transitions; enrollment shares are College Transitions estimates.
At UCLA, 13.0 percent of sections enrolled 100 or more students; under the 150-student assumption, those sections hold 47 percent of estimated enrollments. Michigan (33 percent), Texas A&M (32 percent) and UT Austin (30 percent) place roughly a third of their estimated enrollments in sections of 100 or more under the same assumption.
The Median Enrollment by Class Size
Another way to read the same data is to find the class-size band that contains the median estimated course enrollment, the band at which half of a college’s enrollments are in smaller sections and half in larger ones. This gives a more student-weighted view than the section percentage, although it does not describe any individual student’s schedule, and it does not account for how often or how long different sections meet.
Among the 133 private colleges and universities in the group, the median enrollment falls in sections of 20 to 29 students at 62 schools and 10 to 19 students at 26. Among the 100 public institutions, it falls in sections of 50 to 99 at 46 schools and 40 to 49 at 27. Only two public institutions, New College of Florida and St. Mary’s College of Maryland, each with fewer than 2,000 undergraduates, have a median enrollment in sections under 20.
The pattern tracks undergraduate enrollment closely, as Table 2 shows.
| Undergraduate enrollment | Colleges | Reported: sections under 20 (median) | Estimated: enrollments under 20 (median) | Ceiling (median) |
| Under 3,000 | 72 | 68.9% | 48.1% | 60.2% |
| 3,000 to 9,999 | 58 | 55.6% | 28.0% | 41.0% |
| 10,000 to 24,999 | 53 | 39.4% | 14.2% | 22.3% |
| 25,000+ | 49 | 34.5% | 10.7% | 18.3% |
Table 2. Median reported section share, estimated enrollment share and ceiling for sections under 20 students, by undergraduate enrollment. Spelman College is excluded because its enrollment was not available in the IPEDS file used. Sources: Common Data Set 2025-26, item I-3; IPEDS fall 2024 enrollment; College Transitions estimates.
Table 3 lists the reported figure, the estimate and the ceiling for 34 widely known colleges, ordered by the estimate.
| College | Reported: sections under 20 | Estimated: enrollments under 20 | Ceiling | Median enrollment: class-size band |
| Pomona College | 74.3% | 55.3% | 66.7% | 10 to 19 |
| Williams College | 75.4% | 46.0% | 59.2% | 20 to 29 |
| Northwestern University | 76.8% | 38.3% | 54.1% | 30 to 39 |
| University of Chicago | 74.6% | 36.7% | 52.4% | 20 to 29 |
| Yale University | 73.1% | 35.1% | 50.2% | 30 to 39 |
| Duke University | 71.0% | 35.0% | 49.6% | 30 to 39 |
| Georgetown University | 64.4% | 34.3% | 47.6% | 20 to 29 |
| Dartmouth College | 61.3% | 32.4% | 45.2% | 20 to 29 |
| Princeton University | 72.0% | 30.8% | 46.1% | 40 to 49 |
| Emory University | 63.4% | 29.2% | 42.8% | 30 to 39 |
| Tufts University | 59.8% | 29.0% | 41.8% | 30 to 39 |
| University of Notre Dame | 60.2% | 28.6% | 41.3% | 30 to 39 |
| Harvard University | 73.6% | 28.2% | 44.0% | 50 to 99 |
| Stanford University | 69.7% | 27.7% | 42.5% | 50 to 99 |
| Rice University | 61.6% | 27.1% | 40.2% | 30 to 39 |
| University of Southern California | 60.4% | 26.0% | 38.7% | 40 to 49 |
| Massachusetts Institute of Technology | 68.5% | 25.7% | 40.5% | 40 to 49 |
| Vanderbilt University | 57.4% | 25.6% | 38.1% | 30 to 39 |
| New York University | 56.3% | 25.1% | 37.5% | 30 to 39 |
| Boston University | 57.7% | 22.8% | 34.9% | 40 to 49 |
| Columbia University | 58.1% | 21.9% | 34.7% | 50 to 99 |
| University of Pennsylvania | 58.3% | 19.7% | 32.0% | 50 to 99 |
| University of Florida | 50.5% | 18.6% | 29.3% | 40 to 49 |
| Cornell University | 52.9% | 17.8% | 28.8% | 50 to 99 |
| University of Michigan | 51.2% | 17.3% | 28.2% | 50 to 99 |
| University of California, Berkeley | 50.9% | 13.8% | 23.6% | 50 to 99 |
| University of North Carolina at Chapel Hill | 38.2% | 13.5% | 21.7% | 40 to 49 |
| University of Texas at Austin | 40.7% | 12.8% | 21.5% | 50 to 99 |
| University of Wisconsin-Madison | 44.6% | 12.8% | 21.6% | 50 to 99 |
| Ohio State University | 39.4% | 11.9% | 19.9% | 50 to 99 |
| University of California, Los Angeles | 46.2% | 10.7% | 19.2% | 50 to 99 |
| Pennsylvania State University | 32.0% | 9.6% | 16.2% | 50 to 99 |
| Georgia Institute of Technology | 32.7% | 8.3% | 14.6% | 50 to 99 |
| Texas A&M University | 27.4% | 6.9% | 12.1% | 50 to 99 |
Table 3. Selected colleges, fall 2025. Ceiling is the highest enrollment share the reported section counts allow. The last column is the class-size band containing the median estimated course enrollment. Source: Common Data Set 2025-26, item I-3, compiled by College Transitions; enrollment shares are College Transitions estimates.
Large Universities With the Most Small-Class Enrollments
For students who want a large campus, the useful comparison is among large schools. Figure 3 ranks the 102 colleges in the group with 10,000 or more undergraduates by their estimated share of enrollments in sections under 20, with the reported section share alongside.
Figure 3. Reported share of sections under 20 and estimated share of course enrollments in those sections, fall 2025, the 15 highest estimates among colleges with 10,000 or more undergraduates. Sources: Common Data Set 2025-26, item I-3, compiled by College Transitions; IPEDS fall 2024 enrollment; enrollment shares are College Transitions estimates.
Syracuse (26.5 percent), USC (26.0 percent) and NYU (25.1 percent) lead, all three above the 22.9 percent median for the full group. Eleven of the 15 are private. The public universities on the list are IU Indianapolis, Florida State, Temple and Binghamton.
At the other end, the lowest estimates among the 233 colleges belong to San Jose State (5.3 percent), UC Santa Cruz (5.8 percent), UNC Charlotte, Cal Poly Pomona and Cal State LA (each 6.0 percent).
What This Group of Colleges Represents
The 233 colleges are those with a usable 2025-26 Common Data Set in the College Transitions CDS repository, plus Yale and the University of North Carolina at Chapel Hill, whose reports came from their own websites. The group includes many selective and widely known institutions and is not a representative sample of American colleges, so the medians here describe these 233 schools, not higher education as a whole.
Of the 241 repository files for 2025-26, four left the class-size table blank: Carnegie Mellon, Colgate, Johns Hopkins and Michigan State. Six more could not be used because the file was incomplete, repeated the prior year’s figures, came from a different year, or belonged to a different college. At Auburn, Scripps and UC Santa Cruz, the seven size bands do not add up to the total the school reported, so this analysis uses the band counts. Scripps reports only courses taught at Scripps, not courses its students take at the other Claremont Colleges.
Reading a Class-Size Claim
The table below matches the claims families encounter with what each one measures and what to check next.
| The claim | What it measures | What to check |
| “X percent of classes have fewer than 20 students” | Share of fall class sections under 20, with every section counted once regardless of size | How many sections fall in the 50-to-99 and 100+ bands of CDS I-3 |
| “Two out of three classes have fewer than 30 students” | The same section count with a different cutoff | Whether the cutoff is 20 or 30, and the large-section bands |
| “Average class size is X” | Usually the mean section size, which also counts every section once | Whether the school publishes an enrollment-weighted average; if not, expect larger classes in practice |
| “Student-to-faculty ratio of X to 1” | Students per instructional faculty member (CDS I-2), not the size of any class | Treat it as a staffing measure |
| “Small discussion sections with every lecture” | Subsections such as labs and recitations, reported separately in I-3 | The subsection table, and which courses in the intended major use the lecture-plus-section format |
Table 4. Common class-size claims and how to read them.
For a student comparing a small college with a large university, the estimated distribution of course enrollments offers a useful complement to the reported percentage of small sections. Among colleges with fewer than 3,000 undergraduates, the median estimated share of course enrollments in sections under 20 is 48.1 percent; among those with 25,000 or more, it is 10.7 percent.
Then narrow the question to the student’s intended major. Class sizes vary within a campus, and a prospective engineer and a prospective English major at the same university can have very different first two years. Ask each department how large its introductory courses run, whether those courses meet in separate labs, recitations or discussion sections, and when majors move into seminars.
Methodology and sources: Class-size data are from item I-3 of each college’s Common Data Set 2025-26 (fall 2025 undergraduate class sections, excluding subsections), as compiled in the College Transitions CDS repository, with Yale University and the University of North Carolina at Chapel Hill taken from the reports on their institutional research websites. Values were extracted from PDF, Excel and Word files and checked against each table’s reported total; tables stored as images were read by hand. Every college named in this article was checked against its source table. Estimated course enrollments multiply the sections in each band by the band midpoint (5.5, 14.5, 24.5, 34.5, 44.5 and 74.5), with sections of 100 or more counted at 150 in the main estimate and at 100 and 250 in sensitivity tests. Ceilings place sections of 2 to 9 and 10 to 19 at 9 and 19 students and larger sections at the lower limit of their band. Undergraduate enrollment is from the IPEDS fall 2024 enrollment file. Public or private status reflects the institution’s control. UCLA admissions language is from admission.ucla.edu/explore/academics and the UCLA Viewbook 2026, both accessed October 2026.
