Skip to content
Home » Blog » HTML Form Accessibility: The WCAG 2.1 AA Checklist

HTML Form Accessibility: The WCAG 2.1 AA Checklist

  • by





HTML Forms WCAG 2.1 AA Section 508 ARIA Procurement
16 WCAG Criteria
2.1 AA Federal Anchor
30 min Manual Test
70% Auto Tools Miss

Quick Answer

HTML form accessibility passes WCAG 2.1 AA when every input has a programmatic label, every error is announced to a screen reader, every interactive element is reachable by keyboard with a visible focus ring, and the form reflows cleanly at 200% zoom. Sixteen criteria carry most of the load and automated tools miss roughly 70 percent of failures.

If procurement just sent back your VPAT and flagged the signup form, the rest of your deal is sitting on a few specific WCAG criteria. HTML form accessibility is not abstract here. Specifically, the contracting officer is asking whether the form can be filled out by a blind user with a screen reader, a motor-impaired user on a keyboard, and a low-vision user at 200% zoom. Federal guidance at section508.gov backs the underlying standard.

Most failures cluster around six categories: structure, labels, keyboard, errors, mobile, and ARIA. Therefore this checklist walks each one with the WCAG citation, the failure pattern, and the fix. By the end, you will know what to verify before you re-submit the VPAT.

A clean automated scan is the floor. Procurement tests the ceiling: keyboard, screen reader, zoom, real users.

What HTML Form Accessibility Actually Requires

First, the legal anchor is WCAG 2.1 Level AA. Section 508 references it for federal contracting. Similarly, the DOJ Title II rule references it for state and local agencies. As a result, procurement teams test against it whether you sell to government or to enterprise buyers carrying their own accessibility policies.

Out of 56 success criteria at Level A and AA combined, 16 apply directly to interactive forms. WCAG 2.2 adds a few more (drag movement, target size, focus appearance) and is best practice as of 2026. However, Section 508 has not yet adopted 2.2 and the DOJ Title II rule still anchors to 2.1 AA. In practice, build to 2.1 AA and you pass procurement. Pick up 2.2 where you can while already in the code.

Layer 1: Structure, Labels, And Names

Three categories cover the foundational layer. Importantly, get these right and you have already cleared the most common procurement objection. Specifically, WCAG 1.3.1, 3.3.2, 4.1.2, and 2.4.6 fail together when the underlying HTML is sloppy.

Form Element And Fieldset

Use a real <form> element. Then group related inputs (radio buttons, related checkboxes, address blocks) inside <fieldset> with a <legend>. As a result, screen readers announce the legend before each grouped input. WCAG 1.3.1.

Structure

Label Every Input

Every <input>, <select>, and <textarea> needs a programmatic label using <label for="id"> matched to the input’s id. Placeholder text is not a label. As a result, an unlabeled input fails 1.3.1, 3.3.2, and 4.1.2 at once.

Most-Failed

Required Indicators

Mark required fields with the required attribute (not just an asterisk). Screen readers announce “required” when the attribute is present. Furthermore, include text near the form explaining the asterisk for sighted users. WCAG 3.3.2.

Names & Roles

Layer 2: Keyboard, Focus, And Errors

The second layer covers interaction. Notably, anyone who cannot use a mouse navigates by keyboard. Therefore, every interactive element must be reachable, visible, and behave predictably. Meanwhile, error handling has to announce problems clearly enough that a screen reader user can fix them without seeing the screen.

Tab Order And No Trap

Tab through the form from top to bottom. Every input, button, and link should receive focus in visual order. No element should trap focus. Use tabindex="0" only for custom interactive elements, never positive values. WCAG 2.1.1, 2.1.2, 2.4.3.

Keyboard

Visible Focus Ring

The browser’s default focus outline is accessible. If you remove it for design, replace it with something equally visible (3:1 contrast, 2 pixels minimum thickness). Removing focus styles without replacement is the second-most-common keyboard failure. WCAG 2.4.7 and 2.2’s new 2.4.11.

Often Removed

Errors Announced And Specific

When validation fails, the message must identify the field, describe the problem, and suggest a fix. Use aria-describedby to link the message to the input and role="alert" or aria-live="polite" so screen readers announce it. Red text alone fails 3.3.1 and 3.3.3.

Often Broken

Layer 3: Mobile, Contrast, And ARIA

Finally, the third layer covers visual presentation and the ARIA pitfalls that catch developers who learned accessibility from Stack Overflow. As a result, this is where well-meaning devs ship forms that pass every automated check and still fail real users.

Target Size And Reflow

Tap targets need 24 by 24 CSS pixels minimum under WCAG 2.2 (44 by 44 was the 2.1 AAA target and remains best practice). The form must reflow without horizontal scroll at 320px width and at 200% zoom on a 1280px viewport. WCAG 1.4.10 and 2.5.8.

Mobile

Contrast Ratios

Body text needs 4.5:1 against background. Large text (18pt or 14pt bold) needs 3:1. Furthermore, form input borders, focus indicators, and required markers need 3:1 against adjacent colors under WCAG 1.4.11. Most brand-color form designs fail one of these.

Common Brand Fail

Use ARIA Sparingly

The first rule of ARIA: do not use it if a native HTML element does the job. aria-label on an input with a visible <label> creates a conflict. role="button" on a <div> when you could use <button> is an anti-pattern. Most ARIA VPAT failures come from over-application.

Over-Applied

Approved by a Fortune 500 legal team

Ziv, 508 accessibility specialist

Your website. Accessible?

Find what’s wrong (opens in a new tab)

How To Test What You Just Built

Automated tools catch 30 to 40 percent of WCAG failures. Meanwhile, the remaining 60 to 70 percent need a human running through the form. In practice, 30 minutes of manual testing finds more issues than any scanner. Therefore, run these five steps in order before you mark a form done.

1

Unplug Your Mouse

Tab from the top of the page to the submit button. Every interactive element should receive focus in visual order with a visible ring. Then press Enter on the submit button. Importantly, this single test covers WCAG 2.1.1, 2.1.2, 2.4.3, and 2.4.7.

2

Install NVDA And Tab Through Again

NVDA runs on Windows at no cost. VoiceOver is built into macOS. Does each field announce a meaningful label? Are required fields announced as required? Are help text and error messages announced? If NVDA is silent or wrong on any field, that field fails WCAG 4.1.2.

3

Run An Automated Scan

Open axe DevTools or WAVE on the page. Fix every violation. Then run Lighthouse and confirm the accessibility score. By contrast, automated tools are a floor, not a ceiling. As a result, treat a clean axe scan as “no obvious bugs,” not “WCAG compliant.”

4

Trigger Every Error State

Submit the form empty. Then submit with an invalid email. Finally, submit with a missing required field. Each time, the error must identify the field, explain the problem, and suggest a fix. Focus should move to the first error or be announced via a live region.

5

Open The Form On A Real Phone

Then zoom to 200% on desktop and resize the viewport to 320px wide. Does the form reflow without horizontal scroll? Are all tap targets at least 24 by 24 pixels? This test catches 1.4.10 (Reflow), 2.5.5 (Target Size), and 2.5.8 (Target Size Minimum).

Mapping Your Form Pass To A VPAT Row

Once your form clears the checklist, procurement will ask you to document conformance in a VPAT 2.5. Specifically, each WCAG criterion you tested maps to a row in the report. The form-critical rows you most need to fill out clearly are 1.3.1 (Info and Relationships), 2.4.6 (Headings and Labels), 3.3.2 (Labels or Instructions), and 4.1.2 (Name, Role, Value).

Not sure which VPAT edition applies to your buyer? Most SaaS sellers to US enterprise need the WCAG Edition. By contrast, sellers to federal agencies need the US Federal sub-edition. Furthermore, EU procurement requires the EU sub-edition. The decision comes down to two questions: who asked, and what is the product.

Bugs That Pass Automated Tests And Still Fail Real Users

Automated tools have a known coverage gap. Specifically, they check whether the HTML attribute exists, not whether the value makes sense. As a result, certain bugs ship to production with a clean axe report and still get flagged in procurement review.

Misleading Labels

An input with aria-label="email" passes axe. However, if the visible label says “Username or email” and the aria-label says only “email,” screen reader users hear different information than sighted users. WCAG 2.5.3 (Label in Name) fails.

Hidden Bug

Focus Lost On Validation

Form submits, validation fails, page does not move focus to the error. Meanwhile, sighted users see the red message. By contrast, keyboard and screen reader users have no idea anything happened. This is the most common SPA bug and it never appears in automated reports.

SPA Specific

Custom Controls Without Roles

A “select” built from divs and JavaScript. Looks fine with a mouse, has no role="combobox", no keyboard handlers, and no ARIA states. Passes basic axe checks. As a result, fails the keyboard pass, the screen reader pass, and 4.1.2.

Custom UI

Frequently Asked Questions

What WCAG criteria apply specifically to forms?

Sixteen of the 56 Level A and AA criteria apply directly to interactive forms: 1.3.1, 1.4.3, 1.4.10, 1.4.11, 2.1.1, 2.1.2, 2.4.3, 2.4.6, 2.4.7, 2.5.5, 3.3.1, 3.3.2, 3.3.3, 3.3.4, 4.1.2, and 4.1.3. Specifically, the four that procurement tests most often are 1.3.1, 3.3.2, 4.1.2, and 2.4.7. Get those right and the rest usually follow.

Can I use placeholder text instead of a label?

No. Specifically, placeholder disappears the moment the user starts typing, which fails WCAG 3.3.2 (Labels or Instructions). Furthermore, placeholder is not announced consistently across screen readers, which fails 4.1.2. Use a real <label> element. If space is tight, use a floating label pattern that keeps the label visible after typing.

How long should manual testing take per form?

Roughly 30 minutes for a standard signup or contact form. Specifically, 10 minutes for the keyboard pass, 10 minutes for screen reader testing, and 10 minutes for error triggering and mobile zoom. By contrast, complex multi-step forms (checkout, application forms) take 60 to 90 minutes. By the third form, your team will be faster.

Related Guides In This Series

Approved by a Fortune 500 legal team

Ziv, 508 accessibility specialist

Your website. Accessible?

Find what’s wrong (opens in a new tab)