Crawlers and developer surveys disagree about CSS frameworks, and the gap is the story. W3Techs finds Bootstrap on 13.4% of websites and Tailwind on 0.3%. The State of CSS 2026 survey finds native features like :has() at 83.7% among respondents. This page collects the CSS framework statistics from four datasets, each labelled with what it actually counts.
CSS Framework Statistics At A Glance
:has(), the top feature of 46 surveyedWhich CSS Frameworks Do Live Websites Use?
W3Techs fingerprints public pages and reports what it can detect in shipped code. Bootstrap leads every other framework it tracks by a wide margin.
The largest single group is sites with nothing detected at all. Animate sits second, ahead of Foundation, Tailwind and UIkit combined.
| Framework | Share of all websites | Share of sites with a known framework |
|---|---|---|
| None detected | 82.0% | not applicable |
| Bootstrap | 13.4% | 74.9% |
| Animate | 7.5% | 41.9% |
| Foundation | 0.4% | 2.3% |
| Tailwind | 0.3% | 1.8% |
| UIkit | 0.2% | 0.9% |
| Skeleton | 0.1% | 0.5% |
| Materialize | 0.1% | 0.3% |
| Material Design Lite | 0.1% | 0.3% |
Source: W3Techs, usage statistics and market share of CSS frameworks, 15 September 2026. Percentages measure detected code on surveyed websites. W3Techs notes a website may use more than one CSS framework, so rows overlap and cannot be added together.
Bootstrap And Tailwind CSS Statistics Side By Side
Bootstrap appears on 44.7 times as many surveyed sites as Tailwind. Among developers answering Stack Overflow’s 2025 question on recently added tags, Tailwind CSS 4 was reported by 21.8% of 21,329 respondents.
These two numbers answer different questions. One counts code detected on live pages; the other counts what a self-selected group of developers said they used in the past year. Neither is a count of people, and neither converts into the other. The same split shows up across the wider web coding statistics for 2026.
Devographics, which runs the State of CSS survey, wrote in its 2026 CSS frameworks section that the biggest threat to Tailwind’s position may be no framework at all, pointing to LLMs generating ad-hoc design systems and to CSS itself absorbing use cases that once needed a third-party library.
CSS Feature Usage Statistics From State Of CSS 2026
The 2026 edition ran from 15 May to 29 June 2026 and drew 4,902 respondents. It covered 46 features, down from 63 in 2025.
The survey sorts every feature into three exclusive buckets: used it, heard of it, never heard of it. Across all 46 features the mean “used” score was 31%, so the five below sit far above the typical entry.
| Feature | Used it | Heard of it | Never heard of it |
|---|---|---|---|
:has() | 83.7% | 12.3% | 3.8% |
aspect-ratio | 81.3% | 12.3% | 5.9% |
| CSS Nesting | 70.6% | 23.8% | 5.1% |
scroll-behavior | 68.2% | 18.3% | 12.7% |
| Viewport-Relative Units | 59.8% | 24.7% | 15% |
Source: State of CSS 2026, all features question, responses collected 15 May to 29 June 2026. Opt-in survey of 4,902 respondents; percentages describe respondents, not developers generally.
Anchor Positioning recorded the largest single-year usage increase of any tracked feature at 15 points. It also finished first when respondents were asked which features they avoid because of browser support problems, a question answered by 632 people.
Gap decorations and if() were the least-used features in the 2026 set. Gap decorations was also the item most often saved to the survey’s reading list, chosen by 55% of respondents.
Source: State of CSS 2026, features and reading list sections, May to June 2026.
CSS Page Weight Statistics
The HTTP Archive Web Almanac measures transferred bytes per page. Its 2025 edition, published 15 January 2026 from a July 2025 crawl, puts CSS fourth by weight on the median mobile home page.
Images outweigh stylesheets by roughly twelve to one. Median values by resource type are separate medians and do not sum to the median total page weight of 2,559 KB.
| Resource type | Median bytes, mobile home page |
|---|---|
| Images | 911 KB |
| JavaScript | 632 KB |
| Fonts | 122 KB |
| CSS | 77 KB |
| HTML | 22 KB |
Source: HTTP Archive Web Almanac 2025, Page Weight chapter, Figure 14.5, July 2025 crawl.
How Much CSS Does A Page Actually Ship?
Stylesheet weight is skewed rather than evenly spread. The 90th percentile desktop page carries 274 KB against 38 KB at the 25th.
| Percentile | Desktop CSS | Mobile CSS |
|---|---|---|
| 10th | 10 KB | 7 KB |
| 25th | 38 KB | 34 KB |
| 50th | 83 KB | 79 KB |
| 75th | 161 KB | 156 KB |
| 90th | 274 KB | 268 KB |
Source: HTTP Archive Web Almanac 2025, Page Weight chapter, Figure 14.26, July 2025 crawl.
At the median, CSS grew 4 KB year over year, which the chapter authors describe as roughly 4,000 additional characters. At the 90th percentile it added 8 KB.
The median page made 8 CSS requests on both desktop and mobile. On minification, 61% of desktop and 62% of mobile home page loads passed Lighthouse’s Minify CSS audit, each down one point from 2024. A CSS generator that outputs compact declarations, or a build step that handles keyframe animation output, keeps those files smaller.
Source: HTTP Archive Web Almanac 2025, Page Weight chapter, Figures 14.40 and 14.49, July 2025 crawl.
CSS Statistics From The Stack Overflow Developer Survey
Stack Overflow asked which languages respondents had done extensive development work in over the past year. HTML/CSS ranked second, behind JavaScript.
The survey drew more than 49,000 responses from 177 countries. This particular question was answered by 31,771 people.
| Language | Share of question respondents |
|---|---|
| JavaScript | 66% |
| HTML/CSS | 61.9% |
| SQL | 58.6% |
| Python | 57.9% |
| Bash/Shell | 48.7% |
| TypeScript | 43.6% |
Source: Stack Overflow Developer Survey 2025, technology section, all respondents, n = 31,771. Self-selected survey, fielded 2025. Multiple selections allowed, so the column does not total 100%.
Among professional developers the figure was 63%, on a base of 24,759. Among people learning to code it reached 66.6% on a base of 2,564, where only Python scored higher at 71.8%.
Tailwind CSS 4 appeared in a separate question on recently added Stack Overflow tags. It was reported by 21.8% of the 21,329 respondents to that question and carried a 62.1% admired score, meaning users who would use it again. Shadcn/ui reached 8.7% on the same base. Related shifts show up in the JavaScript statistics for 2026.
Source: Stack Overflow Developer Survey 2025, Stack Overflow tags question, n = 21,329.
Is Tailwind CSS Replacing Bootstrap In 2026?
On live sites, no. Bootstrap accounted for 74.9% of detected framework use and 13.4% of all websites on 15 September 2026, against 0.3% for Tailwind.
Among surveyed developers the picture is different. Tailwind CSS 4 reached 21.8% on Stack Overflow’s tags question in 2025, and Devographics describes native CSS and AI-generated styles rather than a rival framework as the pressure on Tailwind in 2026.
What neither dataset shows is migration. Detection counts what is deployed; surveys count what respondents say they used. A change in one does not document a change in the other. Developers reaching for native alternatives can build border radius, box shadow and gradient declarations without a framework, and transition and multi-column layouts now ship in every major engine. Server-side measurement shows similar inertia in the PHP usage statistics.
FAQs
How many websites use Bootstrap in 2026?
W3Techs detected Bootstrap on 13.4% of all websites it surveyed on 15 September 2026, which is 74.9% of sites where it identifies a CSS framework. This measures code on live pages, not developer or user counts.
Is Tailwind CSS more popular than Bootstrap?
It depends on the dataset. W3Techs detected Tailwind on 0.3% of websites against Bootstrap’s 13.4% on 15 September 2026. Stack Overflow’s 2025 survey recorded Tailwind CSS 4 at 21.8% of 21,329 respondents.
What is the most used CSS feature?
:has(), reported as used by 83.7% of State of CSS 2026 respondents, ahead of aspect-ratio at 81.3% and CSS Nesting at 70.6%. The survey is opt-in and drew 4,902 respondents between May and June 2026.
How much CSS does a web page load?
The median mobile home page transferred 77 KB of CSS and the median desktop home page 82 KB, across 8 CSS requests on both, in the HTTP Archive Web Almanac 2025 crawl of July 2025.
Why does Reddit say Tailwind beat Bootstrap when crawler data shows otherwise?
The two sources count different things. W3Techs fingerprints code already shipping on live sites, where Bootstrap reached 13.4%. Developer surveys record what self-selected respondents say they wrote recently, a separate measurement.
Sources
https://w3techs.com/technologies/overview/css_framework
https://2026.stateofcss.com/en-US/features/
https://almanac.httparchive.org/en/2025/page-weight
https://survey.stackoverflow.co/2025/technology
Related statistics: https://fosspost.org/css-statistics/
