The mechanism behind exercise-induced decreases in plasma insulin concentrations was examined in eight healthy young men. In addition, the influence of specific alpha(1)- and alpha(2)-adrenoceptor blockade on glucose kinetics during exercise was studied. To test the hypothesis that exercise-induced decreases in insulin secretion are mediated via alpha(2)-adrenoceptors, all subjects exercised for 60 min on separate occasions under four conditions: with and without alpha(1)-receptor blockade (1 mg prazosin) and with and without or alpha(2)-receptor blockade (15 mg yohimbine). Glucose kinetics were measured using [3-(3)H]glucose. During exercise with alpha(2)-receptor blockade, the insulin concentration initially increased (first 20 min) then decreased, whereas it continually decreased in the corresponding control experiment. The C-peptide concentration did not change during exercise with alpha(2)-receptor blockade but decreased in the control experiment. During exercise with alpha(1)-receptor blockade and corresponding control experiments, insulin and C-peptide levels always decreased. With alpha(1)-receptor blockade, the glucose concentration increased (first 30 min) and then decreased, whereas it slightly decreased in all other experiments. In addition, with alpha(1)-receptor blockade, the glucose rate of appearance (Ra) increased rapidly (because of higher catecholamine concentrations in alpha(1)-receptor blockade versus control) and the glucose rate of disappearance (Rd) was higher compared with control. During exercise with alpha(2)-receptor blockade, the Ra and Rd were always lower compared with control. Therefore, we conclude that exercise-induced decreases in insulin secretion are mediated via alpha(2)-adrenoceptors and that blockade of alpha(1)- and alpha(2)-adrenoceptors during exercise elicits opposite responses in glucose Ra and Rd.