Ophthalmic and neurological complications are frequent findings in patients with AIDS. Little is known about neuroretinal dysfunction in patients with HIV infection. The purpose of this study was to measure and evaluate colour vision in patients with HIV infection or AIDS. Colour contrast sensitivity tests were performed on 75 patients (150 eyes) in different stages of HIV infection. A highly sensitive computer graphics system was used to measure tritan, deutan, and protan colour contrast thresholds. Patients were classified into three clinical groups: (a) asymptomatic HIV infection, (b) lymphadenopathy syndrome or AIDS-related complex, and (c) AIDS. Overall, tritan (p < 0.0001), deutan (p = 0.003), and protan (p = 0.009) colour contrast sensitivities were significantly impaired in patients with HIV infection compared with normal controls. Colour thresholds in patients with asymptomatic HIV infection (mean tritan threshold: 4.33; deutan: 4.41; protan: 3.97) were not impaired compared with normal controls. Colour vision was slightly impaired in patients with lymphadenopathy syndrome or AIDS-related complex (tritan: 6.25 (p < 0.0001); deutan: 4.99 (p = 0.02); protan: 4.45 (p = 0.05)). In patients with AIDS the impairment was even more marked (tritan: 7.66 (p < 0.0001); deutan: 5.15 (p < 0.0009); protan: 4.63 (p = 0.004)). Analysis of covariance controlling for age demonstrated a close association between impairment of tritan colour contrast sensitivity and progression of HIV disease (p < 0.0001). Following Köllner's rule, our study suggests that neuroretinal dysfunction occurs in patients with symptomatic HIV infection or AIDS. This is emphasised by the finding that the relative impairment in tritan vision compared with deutan/protan vision might reflect the difference in the number of cones or receptive fields. Measurement of tritan colour contrast sensitivity appears to be an appropriate and easily applicable method to detect early neuroretinal dysfunction in patients with HIV disease.