T o assess the rate-limiting step in muscle glycogen synthesis in non-insulin-dependent diabetes mellitus (NIDDM), the concentration of glucose-6-phosphate (G6P) was measured by 31P nuclear magnetic resonance (NMR) during a hyperglycemic-hyperinsulinemic clamp. Six subjects with NIDDM and six age weight-matched controls were studied at similar steady-state plasma concentrations of insulin (approximately 450 pmol/liter) and glucose (11 mmol/liter). The concentration of G6P in the gastrocnemius muscle was measured by 31P NMR. Whole-body oxidative and nonoxidative glucose metabolism was determined by the insulin-glucose clamp technique in conjunction with indirect calorimetry. Nonoxidative glucose metabolism which under these conditions is a measure of muscle glycogen synthesis (1990. N. Engl. J. Med. 322:223-228), was 31 +/- 7 mumol/(kg body wt-min) in the normal subjects and 13 +/- 3 mumol/(kg body wt-min) in the NIDDM subjects (P less than 0.05). The concentration of G6P was higher (0.24 +/- 0.02 mmol/kg muscle) in the normal subjects than in the NIDDM subjects (0.17 +/- 0.02, P less than 0.01). Increasing insulin concentrations to insulin 8,500 pmol/liter in four NIDDM subjects restored the glucose uptake rate and G6P concentrations to normal levels. In conclusion, the lower concentration of G6P in the diabetic subjects despite a decreased rate of nonoxidative glucose metabolism is consistent with a defect in muscle glucose transport or phosphorylation reducing the rate of muscle glycogen synthesis.