In 20 patients with untreated non-insulin-dependent diabetes mellitus (NIDDM), there was a positive relationship between fasting plasma glucose (FPG) and glucose production rate, calculated by the isotope dilution technique (r = 0.72, P < 0.001). This suggests that glucose production rate is an important determinant of FPG in untreated NIDDM. Fifteen patients were also studied during therapy with chlorpropamide for 3–6 mo. During therapy, FPG was lower (133 ± 9 vs. 216 ± 20 mg/dl, mean ± SEM; P < 0.001), glucose production was lower (59.5 ± 2.0 vs 77.6 ± 4.9 mg/m2/min; P < 0.005), and there was a significant correlation between the fall in glucose production and the fall in FPG (r = 0.59, P < 0.05). Fasting IRI levels increased in some, but not all, patients during chlorpropamide (untreated 18 ± 2, treated 21 ± 2 μU/ml; P = NS). However, there was a significant relationship between the percent rise in IRI and the fall in glucose production during treatment (r = 0.75, P < 0.001). Patients with a rise in fasting insulin during therapy had a greater fall in glucose production than those whose insulin did not rise (25.4 ± 8.1 vs. 7.8 ± 2.4 mg/m2/min; P < 0.005). When a low-dose insulin infusion was given to approximate the increases of portal venous insulin during therapy, similar falls of glucose production occurred. We conclude that inhibition of endogenous glucose production during chronic chlorpropamide therapy is an important mechanism for the lowering of FPG and that enhanced insulin secretion is the reason for the major part of this inhibition. The small fall in glucose production in those patients whose insulin level did not rise during therapy suggests an additional contribution by some other mechanism.