How the test works
The logic of the test: what each group of plates is for, how it is scored, and what it cannot tell you.
Dots that only color can sort
Each plate is a disc of dots of different sizes and shades. The figure differs from the background only in color: dot size and lightness are mixed on purpose, so they give nothing away.
The colors are chosen so that, for one kind of color blindness, figure and background look the same. Those color pairs are called confusion colors.
- As printed
- Dot sizes only
- Lightness only
- Color only
Red-green color blindness only
The Ishihara test screens for protans and deutans, the red-green types. It has no plates for tritans, who usually pass it. The four kinds are explained on The basics.
Six groups, six jobs
- Plate 1: readable by everyone. It shows the task, and catches people who cannot see the plates at all.
- Plates 2–9: two numbers in the same dots. A red-green color blind person names the other number, which is positive evidence.
- Plates 10–17: a number that red-green color blind people cannot see. The most efficient screening plates.
- Plates 18–21: a number that mostly only red-green color blind people see. Many people with normal vision see it too, so these plates count least.
- Plates 22–25: two digits, one for protans and one for deutans. The only numeral plates that tell the two types apart.
- Plates 26–38: the same ideas as winding lines, for people who cannot read numerals.
| Group | Plates | Finds red-green color blindness | Tells protan from deutan | Finds tritans |
|---|---|---|---|---|
| Demonstration | Plate 1 | no | no | no |
| Transformation plates | Plates 2–9 | yes | no | no |
| Vanishing plates | Plates 10–17 | yes | no | no |
| Hidden-digit plates | Plates 18–21 | yes | no | no |
| Classification plates | Plates 22–25 | yes | yes | no |
| Tracing plates | Plates 26–38 | some plates | some plates | no |
How the booklet scores it
Of plates 1 to 21: 17 or more read normally means normal color vision; 13 or fewer means red-green color blindness; 14 to 16 is rare, and needs further tests.
A short form uses six plates (1, and one from each of 2–5, 6–9, 10–13, 14–17 and 18–21), and all six must be read correctly.
The booklet asks for daylight, the book held at 75 cm, and about 3 seconds per plate.
What the test cannot do
As a screen for red-green color blindness the test is very good: plates 2 to 17 found between 95.5% and 99% of 401 red-green color blind people, depending on how many mistakes count as a fail. The few who pass are the mildest cases.
The classification plates are much weaker: 18% of strong protans and 3% of strong deutans see neither digit, and about two in five mild cases see both.
It does not find tritans, it does not measure how strong a color blindness is, and it is a screen, not a diagnosis. A proper diagnosis uses an anomaloscope.
Light, and screens
The test was designed for printed plates in daylight. Yellowish indoor light makes some deutans pass, and fluorescent tubes are not suitable either.
Screens change colors in their own ways, and so do settings like Night Shift, True Tone and dark-mode filters. The plates on this site are for understanding the test, not for taking it.
From 16 plates to 38
Shinobu Ishihara, a Japanese ophthalmologist, made his first plates in 1916 for the Japanese army. The international edition of 1917 had 16 plates.
It grew to 32 plates in 1936 and to 38 in 1951. The plates on this site come from the 12th edition of 1957. After Ishihara's death in 1963 the inks were re-specified around 1980, and the test is still printed today.
The number of a plate means little without its edition: the 24- and 14-plate books reuse the same designs in a different order.
- 1916The first plates, for the Japanese army
- 1917The international edition: 16 plates
- 193632 plates
- 195138 plates
- 1957The 12th edition: the plates on this site
- 1963Ishihara dies
- around 1980The inks are re-specified
- TodayStill printed