Color vision and web accessibility: practical design beyond color
GDESIGN
DesignPublished: 10 min read

Color vision and web accessibility: practical design beyond color

People do not all perceive color in the same way. A guide from the Japan Ophthalmologists Association gives frequencies of congenital red–green color vision deficiency in Japan of 5% in its male category and 0.2% in its female category. These are Japan-specific figures, not worldwide estimates. This article explains color vision differences and shows how to make web content understandable without relying on color alone.

The Japanese source uses the C, P, D and T labels proposed by Japan’s Color Universal Design Organization (CUDO). Here, we explain those labels alongside the English terms protan, deutan and tritan. They describe differences, not a hierarchy of people’s abilities. Clinical terms are identified as such.

Japanese original published: 2026-05-11

How color vision works: three cone classes

Most people’s retinas contain three classes of cone photoreceptors with different, overlapping spectral sensitivities. The nervous system compares and processes their signals to produce color perception.

  • L cones: relatively sensitive to longer wavelengths.
  • M cones: relatively sensitive to medium wavelengths.
  • S cones: relatively sensitive to shorter wavelengths.

These are not separate red, green and blue detectors. The most common arrangement is called C-type in CUDO’s terminology. In clinical terminology, dichromacy means that two functional cone classes contribute to color vision. Anomalous trichromacy uses three, but with altered spectral sensitivities. These terms describe sensory mechanisms; this design guide is not a diagnostic assessment.

Many protan and deutan differences relate to variation in genes associated with L- and M-cone photopigments on the X chromosome. Chromosome configurations, gene combinations and X-inactivation vary, so a simple XX/XY binary does not describe every person. The male/female percentages below retain the categories used by the Japanese source; they do not classify an individual’s gender identity.

Protan, deutan and tritan: what the labels mean

These are three important groups of color vision differences, not an exhaustive classification of human vision. Protan and deutan groups include both dichromacy and forms of trichromacy with different cone sensitivities.

Scroll horizontally if the table does not fit.

Representative color vision groups
CUDO labelEnglish termRelated cone classPossible difficulty
P-typeProtanLSome reds can appear darker; some red–green combinations can be difficult to distinguish.
D-typeDeutanMSome red–green combinations can be difficult to distinguish.
T-typeTritanSSome blue–green or yellow–pink combinations can be difficult to distinguish.

Other, rare forms include monochromacy, associated with CUDO’s A-type label. A group name alone does not tell you exactly what someone sees or which everyday tasks they find difficult.

CUDO advocates neutral labels for public communication. Japanese ophthalmology still uses clinical diagnostic terminology, and terminology revisions in the 2000s replaced older Japanese names. The clinical distinctions and CUDO’s public-facing labels serve different purposes. The original Japanese source titles remain unchanged in the references.

Simulated hue bands: useful approximations, not someone’s experience

An original HSV hue band and three model-based transformations for protan, deutan and tritan vision.

This figure applies the severity 1.0 matrices from Machado, Oliveira and Fernandes (2009) in linear RGB. The value 1.0 selects the model’s dichromacy endpoint; it is not a rating of an individual’s ability or everyday difficulty. The original band follows HSV hue on a display; it is not a wavelength spectrum. The transformed rows illustrate possible reductions in color differences, not an individual’s actual visual experience.

Figure method: Machado, Oliveira and Fernandes (2009), severity 1.0. The original paper title and DOI are in the references.

Protan and deutan differences

Some red–green, brown–green or blue–purple combinations can become harder to distinguish. Some reds can appear darker with protan vision. The particular colors, lighting, area, saturation, lightness and degree of the difference all matter.

Tritan differences

Possible difficulties include blue–green combinations and combinations of yellow with some pale colors. The combinations involved differ from those associated with protan and deutan vision.

Use simulation to find possible design problems. People within the same group can have different experiences. Never present a filtered screenshot as an exact reproduction of another person’s sight.

Reading Japan’s prevalence figures in context

Japan Ophthalmologists Association figures for congenital red–green color vision deficiency in Japan: male category 5%, female category 0.2%.

The Japan Ophthalmologists Association’s guide reports:

  • Male category in the source: 5%, or 1 in 20.
  • Female category in the source: 0.2%, or 1 in 500.

These describe population frequencies. They do not determine a person’s perception, gender identity or abilities. Frequencies can vary with region, ancestry and study methods. Do not apply these Japanese figures to all countries or try to infer a visitor’s needs from their identity. Provide non-color cues by design.

Figure source: Japan Ophthalmologists Association, 色覚異常を正しく理解するために. The category names reflect the source’s classification.

Where color-only web interfaces fail

The same five color pairs shown unchanged and with three simulation matrices; the differences depend on the actual colors.

Figure method: Machado, Oliveira and Fernandes (2009), severity 1.0, linear RGB. These are model outputs, not reproductions of individual experience.

Red and green status indicators

Green for success and red for error are familiar conventions. But the distinction can become difficult to see in some conditions. An alert that uses color alone may fail to communicate its meaning.

Chart legends

If color is the only way to match a line or bar to its legend, some readers may be unable to identify the series.

Form validation

A red border alone is an unreliable way to identify an error. A pale border can also lack contrast against the background, regardless of color vision.

Heat maps

Some steps in a red–yellow–green scale can be difficult to distinguish. Add lightness differences, values or patterns, and provide a usable textual or tabular alternative where needed.

Other color pairs

Red–brown and green–yellow can also be difficult combinations, depending on the actual values and viewing conditions. Do not infer a person’s perception from a color name or a group label. Add text or shape cues to navigation and categories.

Relevant WCAG 2.2 criteria

WCAG 2.2 includes several relevant success criteria. The table is a starting point, not a complete accessibility audit.

Scroll horizontally if the table does not fit.

Selected WCAG 2.2 requirements concerning color
CriterionLevelRequirement in brief
1.4.1 Use of ColorADo not use color as the only visual way to communicate information or distinguish something meaningful.
1.4.3 Contrast (Minimum)AAText against its background: at least 4.5:1 for ordinary text, or 3:1 for large text, with exceptions.
1.4.11 Non-text ContrastAAVisual information needed to identify controls and states, or understand graphics: at least 3:1 against adjacent colors, with exceptions.

Large text generally means at least 18 pt (24 CSS px), or bold text of at least 14 pt (about 18.67 CSS px). Criterion 1.4.3 includes exceptions for logotypes and incidental text. Criterion 1.4.11 excludes, for example, inactive controls and unmodified user-agent appearances. Read the full criteria when determining applicability.

Use of Color does not prohibit color. It prevents color from carrying meaning without another visual means of understanding it. Passing a contrast calculation alone does not establish that a design meets this requirement.

Japanese context: JIS X 8341-3:2016 is based on WCAG 2.0, not WCAG 2.2. For example, 1.4.11 was not part of WCAG 2.0. Meeting the few criteria covered here does not justify a claim of site-wide conformance.

Practical design: add cues beyond color

Colored status circles alone compared with circles carrying symbols and the labels Success, Warning and Error.

Combine color with another cue, such as a label, symbol, pattern or shape. This improves the information available to readers, but text legibility and programmatic meaning for assistive technology still need separate checks.

Status indicators

  • Success: a green background, a distinct symbol and a “Success” label.
  • Warning: a yellow background, a warning symbol and a “Warning” label.
  • Error: a red background, an error symbol and an “Error” label.

Charts

Use solid, dashed and dotted lines or different marker shapes, and include text labels in the legend. Make sure the non-color cue appears in both the chart and its legend.

Form validation

Pair the border or icon with a text explanation, such as “Enter an email address in the format name@example.com.”

Inline links

A persistent underline provides a non-color cue. In W3C technique G183, the 3:1 ratio compares link text with surrounding non-link text. Text-to-background contrast is a separate requirement; an underline does not remove it. Also make the link’s location apparent on hover and keyboard focus.

Control states

Provide a visible keyboard-focus indicator, such as an outline. Do not rely on a small enlargement alone. Check contrast against the surroundings and whether other content obscures the indicator. Use text or shapes to communicate states such as selected and unselected.

Evaluate actual color values, not “safe” color names

Color Universal Design (CUD) considers diverse ways of seeing when presenting information. The following colors are examples for evaluating a palette, not a certified combination. No hex code alone guarantees readability or distinguishability: background, area, text size, lightness and display conditions matter.

Scroll horizontally if the table does not fit.

Example sRGB colors compared with a white background
Color exampleHexApproximate ratio against white
Orange-red#FF4B003.36:1 — below 4.5:1 for ordinary text.
Blue-green#03AF7A2.83:1 — below 3:1.
Blue#005AFF5.38:1 — above 4.5:1.
Orange#F6AA001.97:1 — below 3:1.

Ratios here are rounded for display. Determine whether a threshold is met using the unrounded value. These numbers compare each color with white; they do not compare the colors with one another.

Combinations that need care

Red–green, red–brown, green–yellow and blue–purple can be difficult when hue carries the entire distinction. Blue–orange, blue–red or purple–yellow may offer useful alternatives, but only when the actual colors have suitable lightness differences and additional cues.

Grayscale is not a complete test. Lightness differences can help, but a readable grayscale image does not prove accessibility for all forms of color vision. Retain labels, patterns, shapes and line styles.

A WCAG contrast checker can help evaluate the actual foreground/background pair. It cannot establish a whole page’s accessibility.

Use simulation as one part of a review

Browser simulation helps find potential problems with color choices. It does not establish what every visitor experiences.

Chrome DevTools

Open the page, then the Rendering panel and Emulate vision deficiencies. Modes include Protanopia, Deuteranopia, Tritanopia and Achromatopsia. The first three model dichromatic conditions; they do not cover every protan, deutan or tritan experience.

Firefox

Use the simulation options in the Accessibility Inspector where available. Check the version and platform you are using: names and availability can change.

macOS

Accessibility settings under Display → Color Filters can adjust the screen for a user. These filters are display accommodations, not a faithful simulation of someone else’s vision.

Figma and other design tools

Plugins can offer approximate previews. Before using one, review its permissions, maintenance and data-handling behavior.

Do not finish with simulation alone. Implement non-color cues, and where possible test important screens with people who have different color vision characteristics.

Using NanToo’s Japanese tools

The Japanese WCAG checker calculates contrast; the palette generator produces candidate palettes using HSL color relationships; the color converter converts color values. These tools currently have Japanese interfaces. Palette generation does not guarantee CUD or WCAG conformance. Check the real background and non-color cues separately.

A repeatable review checklist

Accessible design makes meaning easier to recover from more than one cue. Color, symbols and labels can help readers with different color vision characteristics, and can also be useful under glare or other difficult viewing conditions.

  • Open the actual page and use DevTools simulation to look for possible problems.
  • Check whether status messages and form errors remain understandable without hue differences.
  • Check chart labels, marker shapes, patterns and line styles.
  • Calculate contrast for the foreground/background pairs you actually use.
  • Verify keyboard-focus visibility, text readability and the meaning available to assistive technology.
  • Include users with different needs in testing where possible.

Do not fit every visitor into a single model of vision. Communicate the same information through multiple clear, usable cues.

References and sources

Editorial note

This article was prepared with AI assistance and reviewed by an editor before publication. It may still contain factual errors, interpretation mistakes, or outdated information. Check the cited primary sources or official documentation before making an important decision.

Related articles