The colour vision of 50 diabetic patients was examined with two screening tests, Standard Pseudoisochromatic Plates part 2 (SPP 2) and Farnsworth Panel D 15 (Panel D 15) test and with two diagnostic tests, Nagel anomaloscope and Farnsworth-Munsell 100-hue test. The performance of the diabetic patients in colour vision tests was compared to their performance in colour dependent urine and blood glucose tests. Fourteen of the patients failed the glucose tests, and they failed both of the screening tests as well. The diagnostic tests showed that all of them had a blue-yellow defect and 10 of them also had a red-green defect. The rest of the patients, 36, read the glucose tests correctly, but 17 of them failed the SPP 2 screening test, and 5 failed the Panel D 15 screening test. In diagnostic tests there were 15 patients with normal colour vision, one patient with a red-green defect, 13 patients with a blue-yellow defect, and 7 patients with both a red-green and a blue-yellow defect. The colour vision defect in diabetic patients is most often a blue-yellow defect or a combined blue-yellow and red-green defect. Therefore, the usual pseudoisochromatic plates, e.g. the Ishihara test, are not sufficient in screening because they screen only red-green defects. The screening tests should contain both a red-green and a blue-yellow part.(ABSTRACT TRUNCATED AT 250 WORDS)