In order to discriminate between insulin-dependent and non-insulin-dependent patients, serum C-peptide concentration was determined using antiserum M1230 in the fasting state and 6 min after an i.v. injection of 1 mg glucagon in 215 patients treated with insulin and 53 treated with diet and oral antidiabetics. A patient was considered well controlled without insulin when fasting blood glucose was below 8 mmol/l and when glucosuria was absent. After re-evaluation of therapy in hospital it was found that the majority of patients with a post-stimulatory serum C-peptide concentration above 0.60 pmol/ml appeared to have non-insulin-dependent diabetes mellitus. When fasting C-peptide was used, a great overlap was found between the two treatment groups. During evaluation of therapy in hospital, 6 previously insulin-treated patients could be well treated with diet and tablets and 6 diet- and tablet-treated patients required insulin. The glucagon test seems to be of value in the outpatient clinic to discriminate non-insulin-dependent from insulin-dependent patients.