To determine the effect of short term intensive insulin therapy in non-insulin-dependent diabetes mellitus (NIDDM) we studied 10 patients who had been on maximal doses of sulphonylurea therapy, with a glycosylated haemoglobin value persistently above the normal range. All patients were non-ketonuric, had negative islet cell antibodies, and had been on sulphonylureas for a mean duration of 5.6 yr. Patients were maintained at euglycaemia (plasma glucose 4-7 mmol/l) for 24 hr using an open-loop intravenous insulin regimen, and then underwent a standard 75 g oral glucose tolerance test (OGTT). This was repeated after 3 months of treatment with insulin. Mean fasting plasma glucose and glycosylated haemoglobin were 10.1 mmol/l and 12.2% respectively before, and 7.1 mmol/l (p less than 0.001) and 8.4% (p less than 0.001) after treatment. There was no significant change in body weight. Plasma insulin and C-peptide responses to 75 g OGTT did not change significantly, but the total amount of intravenously infused insulin required for 24-hr euglycaemia fell from a mean value of 138 u before treatment to 87 u (p less than 0.001) at the end of insulin therapy. Remission, with glycosylated haemoglobin in the normal range for more than 3 months after stopping insulin, was observed in 5 out of the 10 patients. All 5 who failed to achieve remission had markedly blunted maximal insulin responses of less than 10 mu/l on both OGTT's. Our study shows that insulin treatment in NIDDM appears to exercise a beneficial effect by lowering insulin resistance. We suggest that this may be of advantage early on in patients with NIDDM in preserving B-cell reserve.