We recently investigated two patients with diabetes and elevated serum prolactin levels in whom no cause of hyperprolactinaemia could be found. For this reason we measured fasting serum prolactin levels in 72 diabetic males and compared the results with those of 63 healthy males and 90 nondiabetic males attending an Impotence Clinic. The diabetic group had significantly higher serum prolactin levels (13.1 +/- 0.9 ng/ml) than the two control groups (9.9 +/- 0.6 ng/ml for normal males and 7.7 +/- 0.3 ng/ml for the non-diabetic impotent group). Eighteen percent of the diabetics studied had serum prolactin levels above the normal range for males (greater than 20 ng/ml). There was no correlation between serum prolactin levels and duration of diabetes, glycosylated haemoglobin level or presence of clinically apparent retinopathy. The correlation between serum prolactin level and fasting plasma glucose was weak though statistically significant (r = 0.26, P less than 0.05).