To determine whether non-insulin-dependent diabetes mellitus (NIDDM) is characterized by day-long hypoinsulinemia, we measured 24-h serum profiles for glucose, insulin, and C-peptide by use of a constant-rate blood-withdrawal technique in diabetic and control subjects fed isocaloric meals. When only lean subjects were considered, diabetic subjects (relative body weight 0.99 ± 0.3) and control subjects (relative body weight 0.95 ± 0.03) had similar 24-h integrated serum insulin concentrations (13.4 ± 2.5 vs. 16.1 ± 2.0 μU/ml, P NS) due to the offsetting effects of increased basal levels and decreased postprandial responses in NIDDM. In contrast, both basal and meal-stimulated insulin levels were decreased in obese NIDDM subjects (relative body weight 1.39 ± 0.07) compared with obese control subjects (relative body weight 1.60 ± 0.08), resulting in a 61% reduction in the 24-h integrated insulin value (18.7 ± 1.5 vs. 48.4 ± 13.7 μU/ml). Thus, the capacity to increase 24-h integrated serum insulin as a function of relative body weight was impaired in NIDDM subjects (r = 0.27, P NS) compared with control subjects (r = .70, P < .01). In contrast, 24-h integrated C-peptide was decreased (P < .01) in both lean (0.92 ± 0.13 pM/ml) and obese (1.52 ± 0.19 pM/ml) NIDDM patients compared with the respective control groups (1.50 ± 0.13 and 3.03 ± 0.44 pM/ml). The molar ratio of 24-h integrated C-peptide to insulin was diminished in lean but not obese NIDDM compared with control subjects. A 3-wk period of intensive insulin therapy led to normalization of the mean 24-h integrated insulin (but not integrated serum C-peptide) value in NIDDM compared with a control group that had an identical mean relative body weight. The 24-h urinary C-peptide measured on the same day as the serum profile was correlated (P < .01) with both the 24-h integrated serum insulin (r = .69) and C-peptide (r = .67) concentrations in control subjects but not in NIDDM subjects (r = .20 and .04, respectively, P NS). Additionally, the urinary clearance of C-peptide was increased in NIDDM (38.1 ± 7.8 vs. 20.4 ± 1.7 ml/min in control subjects, P < .05) and varied with treatment status (26.0 ± 4.6 ml/min after insulin therapy). We conclude that in NIDDM, 1) lean but not obese untreated patients can have normal 24-h integrated serum insulin values due to the combination of high fasting and decreased postprandial levels, 2) 24-h integrated C-peptide values are decreased in both lean and obese patients, 3) intensive therapy normalizes 24-h integrated serum insulin but not C-peptide levels, and 4) urinary C-peptide per se cannot be used as an index of insulin secretion because urinary C-peptide clearance is elevated and varies with treatment status.