It is well established that patients with non-insulin-dependent diabetes mellitus (NIDDM) are resistant to insulin. However, the contribution of hepatic and extrahepatic tissues to insulin resistance remains controversial. The uncertainty may be at least in part due to errors introduced by the unknowing use in previous studies of impure isotopes to measure glucose turnover. To determine hepatic and extrahepatic responses to insulin in the absence of these errors, steady-state glucose turnover was measured simultaneously with [6-3H]- and [6-14C]glucose during sequential 5- and 4-h infusions of insulin at rates of 0.4 and 10 mU.kg-1.min-1 in diabetic and nondiabetic subjects. At low insulin concentrations, [6-3H]- and [6-14C]glucose gave similar estimates of glucose turnover. Hepatic glucose release was equal to but not below zero in the nondiabetic subjects, but persistent glucose release (P less than 0.001) and decreased glucose uptake (P less than 0.001) was observed in the diabetic patients. At high insulin concentrations, both isotopes underestimated glucose turnover during the 1st h after initiation of the high-dose insulin infusion. More time (P less than 0.05) was required to reachieve steady state in NIDDM than nondiabetic subjects. At steady state, [6-3H]- but not [6-14C]glucose systematically underestimated (P less than 0.05) glucose turnover in both groups due to the presence of a tritiated nonglucose contaminant. The percentage of radioactivity in plasma due to tritiated contaminants was linearly related to turnover.(ABSTRACT TRUNCATED AT 250 WORDS)