The effect of sorbinil (200 mg daily for 4 weeks) was examined in 13 patients, mean age 59.7 years (range 42-72 years), with symptomatic diabetic neuropathy of mean duration 6 year (range 1-18 years). In this double-blind, placebo-controlled crossover trial, studies were made of motor, sensory and autonomic nerve function, severity of painful symptoms and duration of sleep. One patient was withdrawn because of an adverse reaction to sorbinil. In the other 12, constant mean values for glycosylated haemoglobin A1 between 11% and 12% indicated stable though not ideal diabetic control throughout the study. Example values for nerve conduction velocity on placebo and active treatment were: 44.3 +/- 5.9 and 44.8 +/- 5.1 metres/s (mean +/- SD) for median motor nerve, 38.4 +/- 8.2 and 37.2 +/- 7.7 metres/s for median sensory nerve. Thus there was no significant effect of sorbinil on conduction velocity in these or any other of the motor and sensory nerves tested. Abnormal autonomic function was not improved by sorbinil. Subjective pain scores on a 10 cm visual analogue scale were 4.2 +/- 2.4 on placebo and 4.3 +/- 2.4 after sorbinil. Duration of sleep on placebo and active treatment was 6.1 +/- 1.6 and 6.2 +/- 1.7 h/night, respectively. We were not able to detect any beneficial effect of sorbinil on painful diabetic neuropathy in our patients.