A method to calculate unbound cortisol from total cortisol (measured by competitive protein binding) and CBG (measured by radial immunodiffusion) based on the binding equilibrium has been evaluated. The calculated results (y) correlated well with those (x) obtained by centrifugal ultrafiltration at 37.degree.C(y = 1.04 x -2.11 ng/ml; r = 0.975; n = 150). The concentration of CBG is similar in normal men (37.7 .+-. 3.5 (SD) .mu.g/ml; n = 12) and women (39.5 .+-. 3.7 (SD) .mu.g/ml; n = 7) and shows no diurnal variation, but marked diurnal variation is observed for total cortisol (193.7 .+-. 35.0 (SD) ng/ml at 08.00 h vs 43.2 .+-. 23.3 (SD) ng/ml at 22.00 h; n = 19) and particularly for unbound cortisol (16.5 .+-. 5.6 (SD) ng/ml at 08.00 h vs 2.3 .+-. 1.8 (SD) ng/ml at 22.00 h; n = 19). The concentration of CBG (89.1 .+-. 11.2 (SD) .mu.g/ml) and of total cortisol (395.6 .+-. 103.3 (SD) ng/ml at 08.00 h; 110.3 .+-. 16.6 (SD) ng/ml at 22.00 h) are clearly elevated in estrogen treated women (n = 11) but unbound cortisol levels (17.2 .+-. 7.7 SD) ng/ml at 08.00 h; 2.5 .+-. 0.5 (SD) ng/ml at 22.00 h) are similar to the control group. The concentration of CBG is significantly decreased in patients with Cushing''s syndrome (33.2 .+-. 5.6 .mu.g/ml; n = 17) and unbound cortisol is relatively more elevated than total cortisol in these patients. In adrenal insufficiently CBG is normal, but total and unbound cortisol are markedly decreased. There is a significant decrease of CBG in hyperthyroidism (35.7 .+-. 5.5 .mu.g/ml; n = 22), in cirrhosis (32.0 .+-. 8.0 .mu.g/ml; n = 14) and in renal disease and a significant increase in patients treated with antiepileptic drugs (47.5 .+-. 6.3 .mu.g/ml; n = 14), but total and unbound cortisol are normal in all these conditions. We conclude that unbound cortisol can be calculated in a simple and reliable way from total cortisol and CBG and permits a better evaluation of adrenal function, particularly in patients with altered CBG concentrations.