We investigated in six low- and six high-insulin responders (LIR and HIR) the effect of dexamethasone (Dex, 15 mg orally during 48 hr) on oral glucose tolerance (OGTT), glucose turnover under basal conditions and during glucose infusion of 2 mg.kg-1.min-1, and insulin response during hyperglycemic clamp. Dex increased fasting glucose more in LIR (P less than 0.05). During OGTT, Dex caused a more prominent increment in glucose in LIR, whereas the increment in insulin was less in LIR (P less than 0.05). After Dex, in three LIR but in no HIR, a diabetic OGTT was observed. Dex significantly increased basal hepatic glucose production (turnover measured with [6-3H]glucose), hepatic total glucose output (turnover measured with [2-3H]glucose), and glucose cycling (hepatic total glucose output--hepatic glucose production) only in LIR. Dex decreased basal glucose metabolic clearance to the same extent in LIR and HIR. Hyperglycemic clamp revealed that Dex induced a significant increase (P less than 0.05) in insulin response only in HIR. Dex effects on insulin release during hyperglycemic clamp were negatively correlated with the glucose area during Dex OGTT (P less than 0.01). Thus, the double tracer method provided a new insight into the pathogenesis of the steroid effect on carbohydrate tolerance. Dex increased basal glycemia more in LIR because only in LIR was glucose production increased. During OGTT, the LIR who were not able to counteract the effects of Dex by an appropriate enhancement in insulin secretion developed a decreased OGTT. The evaluation of insulin response after Dex may thus allow differentiation of the subset of LIR that run an increased risk of non-insulin-dependent diabetes mellitus.