jsdom vs happy-dom: Should You Switch? 1.9x Faster, 2 Gaps

On my 20-file, 60-test Vitest 5.0.3 suite, happy-dom 20.14.5 finished in a median 1.65 s versus 3.11 s for jsdom 30.1.2, about 1.9x faster. But happy-dom failed 2 of my 8 DOM behavior checks, so switch only if your tests stay clear of the gaps listed below.

Short answer: use happy-dom for large suites of simple render-click-assert tests, and keep jsdom 30 for anything touching focus, table APIs or computed CSS. No affiliate links in this post.

What I tested

• Machine: 8 vCPU / 15 GB RAM Linux cloud box, Node 24.21.0, tested 5 Oct 2026 (IST).
• Runner: Vitest 5.0.3 with Vite 8.3.2, default forks pool, environment set in vitest.config.ts.
• Environments: jsdom 30.1.2 (released 4 Oct 2026) and happy-dom 20.14.5 in one project; jsdom 29.1.1 in a second, otherwise identical project.
• Light suite: 20 files × 3 tests (form submit, a 50-item list, a click that flips aria-expanded).
• Heavy suite: 4 files × 2 tests (build a 2,000-row table with createElement and sort it; parse 2,000 rows through innerHTML and read getComputedStyle on 200 cells).
• Timing: three runs each, median of Vitest's reported Duration. Wall clock with process start is listed further down.
• One surprise: my first heavy test sorted rows with tbody.rows. Under happy-dom it crashed with TypeError: undefined is not iterable. I rewrote the timing test to use querySelectorAll('tr') so both environments run the same work, and kept tbody.rows as a separate compatibility check.
• Not tested: React Testing Library or user-event, memory use, Vitest browser mode, Windows or macOS, a large production suite.

Is happy-dom faster than jsdom in Vitest?

Suite	jsdom 30.1.2	happy-dom 20.14.5	jsdom 29.1.1
Light: 20 files / 60 tests	3.11 s (3.07 / 3.11 / 3.14)	1.65 s (1.54 / 1.65 / 1.70)	2.93 s (3.11 / 2.93 / 2.64)
Heavy: 4 files / 8 tests	1.81 s (1.81 / 1.94 / 1.79)	1.33 s (1.34 / 1.33 / 1.27)	2.55 s (2.50 / 2.55 / 2.56)
Light, wall clock median	3.45 s	2.00 s	3.25 s
Heavy, wall clock median	2.14 s	1.68 s	2.92 s

On the light suite, Vitest said 90–92% of jsdom's time was environment setup, against 82–83% for happy-dom. The win there is per-file startup, so it grows with the number of test files, not the number of assertions. On the heavy suite the gap narrowed to about 1.36x.

jsdom 30 itself is a real upgrade for DOM-heavy tests: the heavy suite went from 2.55 s on 29.1.1 to 1.81 s on 30.1.2, about 29% faster. That matches the 30.1.0 release notes on faster DOM construction and the 30.1.2 fix for slow large-tree builds. The light suite showed no gain (3.11 s vs 2.93 s, and 29.1.1's runs spread from 2.64 s to 3.11 s).

What breaks when you switch to happy-dom?

Check	jsdom 30.1.2	happy-dom 20.14.5	jsdom 29.1.1
C1 getComputedStyle reads a <style> color	pass	pass	pass
C2 CSS.escape('a.b')	pass	pass	fail (CSS undefined)
C3 .card:has(img) selector	pass	pass	pass
C4 empty required input checkValidity() is false	pass	pass	pass
C5 focus() skips a display:none input	pass	fail	fail
C6 width: 2em at 10px font computes to 20px	pass	pass	fail (2em)
C7 getBoundingClientRect().width is 0 (no layout)	0	0	0
C8 tbody.rows and tr.cells	pass	fail	pass
Total	8 / 8	6 / 8	5 / 8

The happy-dom C8 failure is narrow: table.rows and tr.cells worked in a direct check, but tbody.rows (on HTMLTableSectionElement) returned undefined. The C5 failure matters more for accessibility tests: happy-dom moved focus onto a hidden input, so a test asserting "focus lands on the first visible field" can pass when it should fail.

C7 is a reminder for both: neither environment does layout. If a test depends on real sizes, scroll positions or geometry-based IntersectionObserver, use Vitest browser mode or Playwright. I covered Playwright's newest test-runner change in the Playwright 1.63 post.

Does jsdom 30 need a newer Node?

Yes. jsdom 30 requires Node ^22.22.2 || ^24.15.0 || >=26.0.0. On Node 22.20.0, npm i jsdom@30.1.2 printed EBADENGINE warnings (first for its @asamuzakjp/css-color 7.1.3 dependency). My compat file still passed 8/8 there, but that is unsupported. happy-dom 20.14.5 only asks for Node 20+. If you cannot move Node yet, jsdom 29.1.1 or happy-dom are your options; the Node 26 LTS schedule post covers the upgrade timing.

How do I switch without a big-bang change?

1. Keep jsdom as the default environment in vitest.config.ts.
2. Add // @vitest-environment happy-dom at the top of the files that only render, click and read text. Those gain the most, because setup dominates them.
3. Run the whole suite under both environments in CI for a week (environment from an env var works, as in my config) and compare failures.
4. Grep your tests for .focus(, tBodies, .rows and getComputedStyle before you move a file. Those were the risky areas here.
5. If you are already on Vite 8.3.2, no other config change was needed for either environment.

Should you switch from jsdom to happy-dom?

Situation	My pick
Hundreds of small component tests (render, click, assert text)	happy-dom, after checking the gaps above
Tests for focus management, table APIs or computed CSS	jsdom 30
Stuck on Node below 22.22.2	jsdom 29.1.1 or happy-dom
Tests that need real layout or geometry	Neither: browser mode or Playwright

Bottom line: happy-dom cut my light suite from 3.11 s to 1.65 s, and that saving scales with file count. It also got 2 of 8 behavior checks wrong, and one of them (focus on hidden inputs) can hide real accessibility bugs. Who should not switch: teams whose tests exist mainly to catch focus and keyboard regressions.

How this was made: I ran every command above on the ShopperCove test box and kept the raw output; the write-up was drafted with AI help and checked against that output.

Sources

• https://github.com/jsdom/jsdom/releases
• https://github.com/jsdom/jsdom/releases/tag/v30.1.2
• https://github.com/capricorn86/happy-dom
• https://github.com/capricorn86/happy-dom/releases
• https://vitest.dev/guide/environment
• https://www.npmjs.com/package/jsdom
• https://www.npmjs.com/package/happy-dom
• https://www.npmjs.com/package/vitest

Related

• https://www.shoppercove.com/blog/playwright-1-63-test-locks-october-2026
• https://www.shoppercove.com/blog/vite-8-3-2-renderbuilturl-bundled-dev-sourcemaps-october-2026
• https://www.shoppercove.com/blog/vite-plus-1-0-unified-toolchain-october-2026
• https://www.shoppercove.com/blog/nodejs-26-lts-october-2026-schedule-change
• https://www.shoppercove.com/blog/react-19-3-view-transitions-fragment-refs-2026
• https://www.shoppercove.com/blog/chrome-155-stable-jpeg-xl-css-changes-october-2026
• https://www.shoppercove.com/blog/pnpm-12-9-1-rust-rewrite-wasm-split-october-2026
• https://www.shoppercove.com/blog/oxlint-1-87-react-suggestions-a11y-fixes-october-2026