We investigated the effects of caffeine ingestion on skeletal muscle glucose uptake, glycogen synthase (GS) activity, and insulin signaling intermediates during a 100-min euglycemic-hyperinsulinemic (100 microU/ml) clamp. On two occasions, seven men performed 1-h one-legged knee extensor exercise at 3 h before the clamp. Caffeine (5 mg/kg) or placebo was administered in a randomized, double-blind fashion 1 h before the clamp. During the clamp, whole-body glucose disposal was reduced (P < 0.05) in caffeine (37.5 +/- 3.1 micromol x min(-1) x kg(-1)) vs. placebo (54.1 +/- 2.9 micromol x min(-1) x kg(-1)). In accordance, the total area under the curve over 100 min (AUC(0--100 min)) for insulin-stimulated glucose uptake in caffeine was reduced (P < 0.05) by approximately 50% in rested and exercised muscle. Caffeine also reduced (P < 0.05) GS activity before and during insulin infusion in both legs. Exercise increased insulin sensitivity of leg glucose uptake in both caffeine and placebo. Insulin increased insulin receptor tyrosine kinase (IRTK), insulin receptor substrate 1-associated phosphatidylinositol (PI) 3-kinase activities, and Ser(473) phosphorylation of protein kinase B (PKB)/Akt significantly but similarly in rested and exercised legs. Furthermore, insulin significantly decreased glycogen synthase kinase-3alpha (GSK-3alpha) activity equally in both legs. Caffeine did not alter insulin signaling in either leg. Plasma epinephrine and muscle cAMP concentrations were increased in caffeine. We conclude that 1) caffeine impairs insulin-stimulated glucose uptake and GS activity in rested and exercised human skeletal muscle; 2) caffeine-induced impairment of insulin-stimulated muscle glucose uptake and downregulation of GS activity are not accompanied by alterations in IRTK, PI 3-kinase, PKB/Akt, or GSK-3alpha but may be associated with increases in epinephrine and intramuscular cAMP concentrations; and 3) exercise reduces the detrimental effects of caffeine on insulin action in muscle.