How do you make a chart readable for colour-blind readers?

Who is affected and what they lose

Red–green deficiencies are by far the most common. Okabe and Ito give about 8% of Caucasian men, 5% of Asian men and 4% of African men; Crameri and colleagues cite about 8% of men and 0.5% of women worldwide. Most are anomalous trichromats, who see reduced rather than absent red–green differences; a smaller share are dichromats (protanopes and deuteranopes) who lack one cone type. Tritan deficiencies, which affect blue–yellow discrimination, are rare. What all of these remove is a dimension of hue, not lightness. That single fact drives every technique on this page: two colours that differ in lightness remain different for every type, while two colours of equal lightness that differ only along the red–green axis may merge.

What simulation shows, and what it does not

The site's palette checkers simulate dichromacy with the Viénot, Brettel and Mollon (1999) method: colours are converted to linear RGB and projected onto the plane a dichromat can see. That shows the worst case for a full dichromat under standard viewing conditions. It does not show any individual's vision, over-states the loss for the larger group of anomalous trichromats, and cannot account for mark size — small dots and thin lines are harder to tell apart than large areas of the same colours. Treat a simulation result as a screen: if a pair collapses under simulation, fix it; if it survives, still add a second cue, because real charts are read at small sizes on uncalibrated screens.

The Okabe–Ito palette under test

The Okabe–Ito palette is the most widely recommended starting point, and the table shows why and where it still needs help. Under simulated protanopia and deuteranopia, only two of its 21 pairs fall below ΔE00 15, and none below 10, because the palette mixes light, mid and dark colours. Under simulated tritanopia four pairs come closer, notably orange with reddish purple and the two blues with bluish green. In greyscale, orange and sky blue become nearly identical, and several mid-tones crowd together. The lesson is not that the palette is flawed but that no seven-colour palette survives every condition: assign the weakest pairs to series that never sit next to each other, or give them different marker shapes.

Redundant encoding that actually works

Direct labels are the most reliable redundancy: a series name at the end of each line removes the legend lookup altogether, which is also what Okabe and Ito recommend. For scatter plots, vary marker shape (circle, triangle, square) as well as colour, and keep shapes to about five, since beyond that they too become hard to tell apart. For line charts, solid, dashed and dotted styles help for two or three series. For bar charts, order the legend in the same order as the bars and label bars directly where space allows. Patterns and hatching work in print but can create visual noise on screen; use them sparingly for areas such as stacked bars. And make hover or focus states show the series name, so interactive charts never rely on colour lookup.

The data

What goes wrong, and what to do

How to check it yourself

1. Run the palette in the data-visualisation palette checker and read the worst pair under each condition.
2. Preview a finished chart through the colour-vision deficiency simulator, not just the swatches.
3. Remove colour entirely (greyscale) and check that every series can still be identified from labels, shapes or position.

What the evidence says

Each statement is labelled by kind — established fact, a standard’s requirement, observed market data, a convention, or Colourwise’s own interpretation or analysis — with the strength of the evidence behind it.

• FactStrong evidence

  Crameri, Shephard and Heron cite a worldwide prevalence of colour-vision deficiency of about 8% of men and 0.5% of women, higher in populations with a larger share of people of northern European descent.

  Caveat: Estimates vary with population and test method.

  Source: The misuse of colour in science communication; Colour Blindness / Colour Vision Deficiency (patient guidance)

• FactStrong evidence

  Viénot, Brettel and Mollon published a method for simulating how colours on a display appear to dichromats, by projecting linear RGB values onto the colours a dichromat can distinguish.

  Source: Digital video colourmaps for checking the legibility of displays by dichromats

• ConventionModerate evidence

  Okabe and Ito recommend combining colour with differences in shape, position, line type or pattern, and labelling chart elements directly rather than relying on a colour-keyed legend.

  Source: Color Universal Design (CUD) — How to make figures and presentations that are friendly to colorblind people

• Colourwise analysisStrong evidence

  In greyscale the Okabe–Ito orange (#E69F00) and sky blue (#56B4E9) differ by less than 1 ΔE00, although they remain distinct under simulated protanopia and deuteranopia.

  Based on: Computed by Colourwise with the palette checker's analysis; see the pair table.

  Source: Color Universal Design (CUD) — How to make figures and presentations that are friendly to colorblind people; Digital video colourmaps for checking the legibility of displays by dichromats

Related questions

Which colour-vision deficiency should I design for?

Deuteranomaly and the other red–green types first, because they are most common; then check tritanopia and greyscale. A palette that differs in lightness helps with all of them at once.

Are colour-blind simulators accurate?

They are good models of full dichromacy under standard conditions, and the right tool for catching bad pairs. They do not show any individual's vision and over-state the loss for milder, more common deficiencies.

Is it enough to use a colour-blind-safe palette?

It removes the worst failures, but small marks and printing still make some pairs hard to tell apart. Direct labels or shape differences are what make a chart reliably readable.

Related topics

Go further

Sources

• The misuse of colour in science communication — Crameri, Shephard and Heron — Nature Communications 11, 5444 (2020) (Academic)
• Colour Blindness / Colour Vision Deficiency (patient guidance) — National Eye Institute, US National Institutes of Health (Museum or institution)
• Digital video colourmaps for checking the legibility of displays by dichromats — Viénot, Brettel and Mollon — Color Research & Application 24(4), 243–252 (1999) (Primary research)
• Color Universal Design (CUD) — How to make figures and presentations that are friendly to colorblind people — Masataka Okabe and Kei Ito (hosted by the University of Cologne) (Academic)
• CIE (International Commission on Illumination) publications — CIE (Official standard or government)

Reviewed 29 September 2026. Colourwise summarises its sources in its own words and does not reproduce standards text or proprietary colour data. Spotted an error? Tell us.