Subtle alterations in neurological function are often difficult to identify and even harder to quantitate. The identification of a neurotoxic state existing before overt behavioral changes occur has eluded quantification. It was hypothesized that a challenging signal-detection procedure would be used to assess neurological function of dialysis patients and other subjects, the degree of uremic toxicity occuring during an interdialytic interval, and the effects of neuroactive drugs. A vigilance task demanding the detection of an irregularly flashing light from a matrix of regularly flashing lights was administered to 3 groups of 15 men: patients undergoing maintenance hemodialysis, patients with chronic illness and no kidney disease, and healthy subjects. The procedure was found to yield a reliable measure; average test-retest correlation was 0.93, which differentiated not only within the hemodialysis cycle (p less than 0.001), between groups (p less than 0.001), but was also related to the recency of neuroactive drugs ingested (p less than 0.01).