1 The aims of this study were to assess the relative beta1/beta2 selectivity of the antagonist and partial agonist activity (PAA) of celiprolol in man. 2 Eight normal males received single oral doses of celiprolol 200 mg (C200), 400 mg (C400) and 800 mg (C800); atenolol 50 mg (A50), 100 mg (A100) and 200 mg (A200); nadolol 40 mg (N40) and placebo (PL), administered in a single-blind, randomised crossover design. 3 At rest, in the presence of low levels of circulating adrenaline and noradrenergic tone, a low dose of celiprolol (C200) showed evidence of beta1-PAA by significant increases in systolic blood pressure and resting heart rate. At higher doses (C400, C800), beta2-PAA became evident by a significant increase in postural finger tremor, whereas C200 had no effect. 4 In the presence of a beta1-adrenoceptor agonist, as assessed by reduction of exercise tachycardia, increasing doses of celiprolol produced signficantly less beta1-adrenoceptor blockade compared with atenolol. Furthermore, there was no increase in beta1-adrenoceptor blockade beyond C400. 5 In the presence of a beta2-adrenoceptor agonist, as assessed by blunting of terbutaline-induced chronotropic, hypokalaemic and finger tremor responses, celiprolol exhibited less beta2-adrenoceptor blockade than comparable doses of atenolol used in clinical practice. 6 Exercise hyperkalaemia was blunted significantly by C400 and C800 in comparison with all doses of atenolol and nadolol. 7 Thus, the partial agonist and antagonist activity of celiprolol is relatively beta1-selective at lower doses, with dose-related loss of selectivity for both beta-adrenoceptor blockade and PAA. Furthermore celiprolol exhibited less beta1- and beta2-adrenoceptor antagonism in comparison with atenolol at each of the dose increments used. Attenuation of exercise hyperkalaemia appears to be a further method of evaluating beta2-PAA in man.