We compared glycemic control achieved on an outpatient basic with three insulin regimens in 10 patients with insulin-dependent diabetes mellitus. The regimens studied included: (1) intensified conventional therapy with twice-daily regular and lente insulin; (2) intensified conventional therapy with long-acting ultralente insulin plus multiple preprandial injections of regular insulin; (3) continuous subcutaneous insulin infusion. Each treatment period was two months long. At the beginning of the study and the close of each study period, patients were hospitalized for a 48-hour evaluation of glycemic control. Each new insulin regimen was begun after discharge, with the dosage adjusted using preplanned algorithms, patient self-monitoring of blood glucose and defined blood glucose targets. Glycemic control markedly improved on all three treatment regimens, to a comparable degree, as assessed by mean plasma glucose level, mean amplitude of glycemic excursions, M value (an index of glycemic lability), urinary glucose excretion and glycosylated hemoglobin level.