20 patients with non insulin dependent diabetes mellitus (NIDDM), and within 20% of their ideal body weight were studied. They had failed to achieve adequate diabetic control following 3 months of dietary therapy. They were randomly allocated to insulin or sulphonylurea therapy for 3 months and then "crossed over" for the same period of time. Patients were maintained at euglycaemia (plasma glucose 4-7 mmol/l) for 24 hr using an open-loop intravenous insulin infusion, and then underwent a standard 75 gm oral glucose tolerance test (OGTT) following each mode of therapy. Mean glycosylated haemoglobin and preprandial blood glucose were 8.7% and 7.7 mmol/l respectively after sulphonylurea, and 7.8% (p less than 0.05) and 6.6 mmol/l (p less than 0.05) after insulin therapy. There was no significant difference in change in body weight. Following a 75 gm OGTT mean plasma insulin at 1/2 hr and 1 hr was 14.0 mu/l and 16.5 mu/l after sulphonylurea, and 23.4 mu/l (p less than 0.05) and 22.1 mu/l (p less than 0.05) after insulin therapy. Plasma C-peptide responses were also improved at 1/2, 1 and 1 1/2 hr after a period of insulin therapy being 0.61 nmol/l, 0.65 nmol/l and 0.59 nmol/l respectively. After sulphonylurea therapy comparable plasma C-peptide responses were 0.31, 0.41 and 0.37 nmol/l respectively (p less than 0.05). There was no significant difference in the total amount of intravenous insulin required for 24 hr euglycaemia. Our study shows that short term insulin therapy in patients with NIDDM who had failed on diet alone has advantages over sulphonylurea therapy in that it achieves better diabetic control and an improved B-cell response to glucose stimulation.(ABSTRACT TRUNCATED AT 250 WORDS)