While caffeine impedes insulin-mediated glucose disposal in humans, its effect on endo-genous glucose production (EGP) remains unknown. In addition, the mechanism involved in these effects is unclear, but may be due to the accompanying increase in adrenaline concentration. We studied the effect of caffeine on EGP and glucose infusion rates (GIR), and whether or not adrenaline can account for all of caffeine's effects. Subjects completed three isoglycaemic-hyperinsulinaemic clamps (with 3-[³H]glucose infusion) 30 min after ingesting: (1) placebo capsules (n= 12); (2) caffeine capsules (5 mg kg^-1) (n= 12); and either (3) placebo plus a high-dose adrenaline infusion (HAdr; adrenaline concentration, 1.2 nM; n= 8) or (4) placebo plus a low-dose adrenaline infusion (LAdr; adrenaline concentration, 0.75 nM; n= 6). With caffeine, adrenaline increased to 0.6 nM but no effect on EGP was observed. While caffeine and HAdr decreased GIR by 13 (P < 0.05) and 34% (P < 0.05) versus the placebo, respectively, LAdr did not result in a significant reduction (5%) in GIR versus the placebo. Due to the fact that both caffeine and LAdr resulted in similar adrenaline concentrations, but resulted in different decreases in GIR, it is concluded that adrenaline alone does not account for the effects of caffeine and additional mechanisms must be involved.