There are no articles in the international psychiatric literature reporting subclinical visual system disorders in depressed patients, although a disturbance of circadian rhythms is one of the prominent theories of the etiopathogenesis of depression. Fifty patients aged 21-60 years suffering from major depression according to DSM-IV criteria, and 20 controls took part in the study. Diagnosis was obtained with the aid of the Schedules for Clinical Assessment in Neuropsychiatry version 2.0. Psychometric assessment included the Hamilton Depression Rating Scale, the Hamilton Anxiety Scale, the 1965 and 1971 Newcastle Scales, and the Diagnostic Melancholia Scale. All subjects had normal electroretinographic and flash-visual evoked potential (VEP) recordings. Pattern-reversed VEPs (PR-VEPs) were recorded from each eye separately. Three-way analysis of covariance, Student's t-test and Pearson product-moment correlation coefficients were used for the analysis. All recordings were within the normal range. N80 and P100 latency were significantly shorter in atypical and significantly longer in melancholic patients. There was a positive correlation between N80 and P100 latency and age of onset and melancholic indices, and a negative correlation with the presence and the number of life events precipitating onset. The results of the current study suggest that PR-VEPs are consistent with other biological data supporting the atypical-melancholic distinction. The most important finding was the strong negative relationship between PR-VEP latency and stressful life events. The current study also provided data inconsistent with the hyperarousal theory and in support of an arousal dysregulation hypothesis for major depression.