The therapeutic effect of tolbutamide (1.5 g daily) in a random sample of patients with non-insulin dependent diabetes mellitus (NIDDM), was studied in a controlled, double-blind cross-over trial of 13 women and 6 men, aged 40-65 years and of 85-155% ideal body weight. The trial comprised C-peptide determinations during a standard carbohydrate rich meal followed by four periods of 3 months in which alternating tolbutamide and placebo were given. From the beginning to the end of the treatment periods fasting blood glucose was reduced from 11.9 +/- 1.1 (mean +/- SEM) to 10.0 +/- 0.8 mmol/l (P less than 0.025), glycohaemoglobin from 12.8% +/- 0.7 to 11.3% +/- 0.5 (P less than 0.02) with a close correlation between fasting blood glucose and glycohaemoglobin (r = 0.87, P less than 0.001). The observations during the first 3 months of study was not included in the calculations. Fasting C-peptide and fasting insulin concentrations were not significantly altered by tolbutamide treatment. The effect of tolbutamide was inversely correlated to the C-peptide response to the standard test meal at the start of the trial (r = 0.76, P less than 0.01), so that patients with the most pronounced beta-cell failure had the greatest therapeutical effect. The beta-cell response to the test meal could not identify patients, whose fasting blood glucose would be normalized by tolbutamide treatment.