To investigate the effects on 24-h insulin and glucose profiles of two- and four-dose insulin regimens in patients with non-insulin-dependent diabetes mellitus (NIDDM) who have failed oral agent therapy. Ten patients with NIDDM and 10 matched nondiabetic control subjects took part in the study, a randomized crossover trial with 8-wk treatment periods. We determined 24-h profiles of blood glucose and free insulin in control subjects and patients when they were taking oral agents and at the end of each treatment period. The two-dose regimen was mixed (15% regular, 85% NPH) insulin given before breakfast and dinner, and the four-dose regimen was regular insulin given before meals and NPH given at 2200. Patients taking oral agents had higher mean blood glucose than control subjects (mean +/- SE 12.6 +/- 0.6 vs. 4.5 +/- 0.1 mM, P less than 0.0001) and similar 24-h insulin concentrations but lower postprandial insulin concentrations (at breakfast, 87.6 +/- 13.8 vs. 228.6 +/- 29.4 pM, P less than 0.01). Mean 24-h insulin concentrations were the same on two- and four-dose regimens, and both regimens caused basal hyperinsulinemia. Glycemic control also improved. Postprandial insulin peaks were higher at lunch and dinner on the four-dose regimen, and postprandial blood glucose was lower. Fasting proinsulin was elevated in patients compared with the control subjects (25.1 +/- 4.7 pM vs. 5.0 +/- 0.9 pM, P less than 0.001) and was suppressed to normal during insulin treatment. Two- and four-dose insulin regimens can achieve similar glycemic control. Both regimens cause basal hyperinsulinemia but normalize the hyperproinsulinemia observed during failure of oral agent therapy. Four-dose insulin regimens offer few advantages to the glycemic control achieved with two-dose regimens but may be more physiological.