Subjects with NIDDM have increased plasma proinsulin concentrations, compared with nondiabetic subjects, both in absolute terms and as a proportion of circulating insulin-like molecules. It remains uncertain whether this reflects a primary beta-cell defect in proinsulin processing or is secondary to hyperglycemia. We addressed this question by assessing the effects of reducing hyperglycemia on relative hyperproinsulinemia in subjects with NIDDM. Eight subjects with NIDDM underwent three 8-week periods in a randomized crossover design of therapy with diet alone, sulfonylurea (gliclazide), or insulin (ultralente). The effects on beta-cell peptide concentrations were assessed 1) fasting, 2) in response to hyperglycemic clamping, and 3) in response to an injection of the nonglucose secretogogue arginine and compared with measurements in seven nondiabetic control subjects. Both sulfonylurea and insulin therapy substantially reduced fasting plasma glucose and glycosylated hemoglobin (HbA1e) concentrations, compared with diet therapy alone. The diabetic subjects on diet therapy had relative hyperproinsulinemia, assessed relative to C-peptide concentrations, fasting and in response to hyperglycemic clamping and arginine, compared with control subjects. Neither sulfonylurea nor insulin therapy altered the relative hyperproinsulinemia. Insulin therapy reduced fasting proinsulin concentrations from geometric mean 29.4 (1 SD range, 14.6-59.0) pmol/l on diet therapy to 18.7 (7.3-48.1) pmol/l (P < 0.05). A similar trend was evident with fasting C-peptide concentrations with a reduction from 0.9 (0.6-1.4) nmol/l on diet therapy to 0.6 (0.4-0.9) nmol/l (P = 0.06), so that the relative hyperproinsulinemia, assessed as the ratio of fasting proinsulin to C-peptide, was unchanged by insulin. Similarly, insulin therapy failed to reduce the ratio of proinsulin to C-peptide concentrations in response to a hyperglycemic clamp and in the acute incremental response to arginine. Failure to improve the relative hyperproinsulinemia of NIDDM, despite significant reduction of hyperglycemia with exogenous insulin therapy, supports the hypothesis that relative hyperproinsulinemia in NIDDM is a reflection of a primary beta-cell defect rather than being secondary to hyperglycemia.