VPAT Draft DeskStart your VPAT

WCAG 2.2 AA checklist in plain language (all 55 criteria)

All 55 WCAG 2.2 success criteria at Levels A and AA, each rewritten as a question your team can answer about your product. Use it to prepare a VPAT, plan an audit, or brief your engineers.

Updated October 1, 2026 · 12 min read

There are 31 Level A and 24 Level AA criteria. Rows marked NEW IN 2.2 were added in WCAG 2.2; skip them if you are reporting against WCAG 2.1, and add 4.1.1 Parsing instead. The questions are a plain-language summary, not the normative text. When a row is borderline, read the full criterion and its “Understanding” document on the W3C site.

1. Perceivable

People can perceive the content: text alternatives, captions, structure, color and contrast, and layout that adapts.

No.CriterionAsk your team
1.1.1Non-text ContentLevel ADo images, icons, and charts have text alternatives that give the same information?
1.2.1Audio-only and Video-only (Prerecorded)Level ADo prerecorded audio-only or video-only clips have a transcript or an equivalent alternative?
1.2.2Captions (Prerecorded)Level ADo prerecorded videos with sound have accurate captions?
1.2.3Audio Description or Media Alternative (Prerecorded)Level ADo prerecorded videos have audio description or a full text alternative?
1.2.4Captions (Live)Level AADo live videos with sound have captions?
1.2.5Audio Description (Prerecorded)Level AADo prerecorded videos have audio description of important visual content?
1.3.1Info and RelationshipsLevel AAre headings, lists, tables, and form labels marked up so assistive technology can tell what they are?
1.3.2Meaningful SequenceLevel ADoes the reading order in the code match the order people see on screen?
1.3.3Sensory CharacteristicsLevel ADo instructions work without relying only on shape, size, position, or sound?
1.3.4OrientationLevel AADoes content work in both portrait and landscape orientation?
1.3.5Identify Input PurposeLevel AADo fields that collect the user’s own details (name, email, address) identify their purpose with autocomplete?
1.4.1Use of ColorLevel AIs color never the only way information is shown, such as status, errors, required fields, or links?
1.4.2Audio ControlLevel AIf audio plays automatically for more than 3 seconds, can people pause it or turn it down?
1.4.3Contrast (Minimum)Level AAIs normal text at least 4.5:1 against its background, and large text at least 3:1?
1.4.4Resize TextLevel AACan text be resized to 200% without losing content or function?
1.4.5Images of TextLevel AAIs text shown as real text rather than images of text, except in logos?
1.4.10ReflowLevel AADoes content reflow at 320 CSS pixels wide without scrolling in two directions?
1.4.11Non-text ContrastLevel AADo controls, focus indicators, and meaningful graphics have at least 3:1 contrast?
1.4.12Text SpacingLevel AADoes content still work when people increase line, letter, word, and paragraph spacing?
1.4.13Content on Hover or FocusLevel AACan tooltips and pop-ups that appear on hover or focus be dismissed, hovered, and kept open?

2. Operable

People can operate the interface: keyboard access, enough time, no seizure risks, clear navigation, and pointer and touch input.

No.CriterionAsk your team
2.1.1KeyboardLevel ACan every function be used with a keyboard alone?
2.1.2No Keyboard TrapLevel ACan keyboard users always move focus out of every component, including dialogs and widgets?
2.1.4Character Key ShortcutsLevel AIf there are single-key shortcuts, can people turn them off or remap them?
2.2.1Timing AdjustableLevel AIf there are time limits, can people turn them off, adjust them, or extend them?
2.2.2Pause, Stop, HideLevel ACan people pause, stop, or hide content that moves, blinks, scrolls, or updates on its own?
2.3.1Three Flashes or Below ThresholdLevel ADoes nothing flash more than three times in any one second?
2.4.1Bypass BlocksLevel ACan people skip repeated blocks such as navigation, with a skip link or landmarks?
2.4.2Page TitledLevel ADoes every page or screen have a title that describes it?
2.4.3Focus OrderLevel ADoes keyboard focus move in an order that keeps meaning and operation intact?
2.4.4Link Purpose (In Context)Level AIs each link’s purpose clear from its text or its immediate context?
2.4.5Multiple WaysLevel AAIs there more than one way to find a page, such as navigation, search, or a sitemap?
2.4.6Headings and LabelsLevel AADo headings and labels describe their topic or purpose?
2.4.7Focus VisibleLevel AAIs keyboard focus always visible?
2.4.11Focus Not Obscured (Minimum)Level AA NEW IN 2.2Is the focused element never completely hidden by sticky headers, footers, or overlays?
2.5.1Pointer GesturesLevel ACan anything done with multi-finger or path-based gestures also be done with a single tap or click?
2.5.2Pointer CancellationLevel ADo actions run on release rather than on press, or can they be aborted or undone?
2.5.3Label in NameLevel ADoes each control’s accessible name include its visible label text?
2.5.4Motion ActuationLevel ACan anything triggered by shaking or tilting also be done through the interface, and be turned off?
2.5.7Dragging MovementsLevel AA NEW IN 2.2Can anything done by dragging also be done with a single click or tap?
2.5.8Target Size (Minimum)Level AA NEW IN 2.2Are click and tap targets at least 24 by 24 CSS pixels, or spaced so they don’t crowd each other?

3. Understandable

People can understand the content and how the interface behaves: language, predictable behavior, and help with input.

No.CriterionAsk your team
3.1.1Language of PageLevel AIs the default language of each page set in the code?
3.1.2Language of PartsLevel AAAre passages in a different language marked with their language?
3.2.1On FocusLevel ADoes moving focus to an element never cause an unexpected change of context?
3.2.2On InputLevel ADoes changing a setting or field never cause an unexpected change of context without warning?
3.2.3Consistent NavigationLevel AADoes repeated navigation appear in the same order across pages?
3.2.4Consistent IdentificationLevel AAAre components with the same function labeled the same way throughout?
3.2.6Consistent HelpLevel A NEW IN 2.2Is help, such as contact details, chat, or a help link, in the same relative place on every page?
3.3.1Error IdentificationLevel AWhen an input error is detected, is the field identified and the error described in text?
3.3.2Labels or InstructionsLevel ADo fields that need input have labels or instructions?
3.3.3Error SuggestionLevel AAWhen an error is detected and a fix is known, is the fix suggested?
3.3.4Error Prevention (Legal, Financial, Data)Level AAFor legal, financial, or data-changing submissions, can people review, correct, or reverse them?
3.3.7Redundant EntryLevel A NEW IN 2.2Is information already entered in a process filled in or selectable, rather than asked for again?
3.3.8Accessible Authentication (Minimum)Level AA NEW IN 2.2Can people sign in without a memory or transcription test, or is an alternative offered?

4. Robust

Content works with current and future assistive technologies: names, roles, values and status messages.

No.CriterionAsk your team
4.1.2Name, Role, ValueLevel ADo custom controls expose their name, role, state, and value to assistive technology?
4.1.3Status MessagesLevel AAAre status messages, such as “Saved” or a results count, announced without moving focus?

From checklist to report

Once your team has answered these questions, you have what you need for a VPAT. For each row, choose a conformance level and write a remark in your own words.

VPAT Draft Desk asks these same questions one at a time and turns your answers into a finished WCAG 2.2 or 2.1 report in PDF and Word.

Start your VPAT

These guides explain common practice for software vendors. They are not legal advice, and VPAT Draft Desk is not affiliated with ITI or the W3C.