We have studied the effects of 3 wk of continuous subcutaneous insulin infusion (CSII) on endogenous insulin secretion and action in a group of 14 type II diabetic subjects with a mean (±SEM) fasting glucose level of 286 ± 17 mg/dl. Normal basal and postprandial glucose levels were achieved during insulin therapy at the expense of marked peripheral hypserinsulinemia. During the week of posttreatment evaluation, the subjects maintained a mean fasting glucose level of 155 ± 11 mg/dl off insulin therapy, indicating a persistent improvement in carbohydrate homeostasis. Adipocyte insulin binding and in vivo insulin dose-response curves for glucose disposal using the euglycemic clamp technique were measured before and after therapy to assess the effect on receptor and post-receptor insulin action. Adipocyte insulin binding did not change. The insulin dose-response curve for overall glucose disposal remained right-shifted compared with age-matched controls, but the mean maximal glucose disposal rate increased by 74% from 160 ± 14 to 278 ± 18 mg/m2/min (P < 0.0005). The effect of insulin treatment on basal hepatic glucose output was also assessed; the mean rate was initially elevated at 159 ± 8 mg/m2/min but fell to 90 ± 5 mg/m2/min in the posttreatment period (P < 0.001), a value similar to that in control subjects. Endogenous insulin secretion was assessed in detail and found to be improved after exogenous insulin therapy. Mean 24-h integrated serum insulin and C-peptide concentrations were increased from 21.377 ± 2766 to 35.584 ± 4549 μU/ml/min (P < 0.01) and from 1653 ± 215 to 2112 ± 188 pmol/ml/min (P < 0.05), respectively, despite lower glycemia. Second-phase insulin response to an intravenous (i.v.) glucose challenge was enhanced from 170 ± 53 to 1022 ± 376 μU/ml/min (P < 0.025), although first-phase response remained minimal. Finally, the mean insulin and C-peptide responses to an i.v. glucagon pulse were unchanged in the posttreatment period, but when glucose levels were increased by exogenous glucose infusion to approximate the levels observed before therapy and the glucagon pulse repeated, responses were markedly enhanced. Simple and multivariate correlation analysis showed that only measures of basal hepatic glucose output and the magnitude of the postbinding defect in the untreated state could be related to the respective fasting glucose levels in individual subjects. We conclude that after 3 wk of intensive insulin therapy, diabetic subjects maintain lower glucose values concomitant with: (1) partial reversal of the postbinding defect in peripheral insulin action, (2) near-normalization of basal hepatic glucose output, and (3) enhanced insulin secretory responses. First-phase insulin response remained minimal and may be a marker for the diabetic state. Correlation analysis could only implicate basal hepatic glucose output and the postbinding defect in the untreated state as direct determinants of the fasting glucose level.