The evaluation of a recently introduced kit (Cortipac (R)) for measuring plasma cortisol by competitive protein binding technique without prior extraction of steroids from plasma, is presented. The blank value was less than 2.5 mug/100 ml, and the lowest measurable value was 2.5 mug/100 ml. The coefficients of variation within- and between-batch determinations were 8.64% and 9.21% respectively. By adding known amounts of cortisol to pre-extracted plasma a linear correlation between calculated and measured cortisol was obtained. In comparison with the double isotope derivate technique, a linear correlation was found. A high degree of cross-reaction was found with nearly all of 11 tested steroids. Due to the considerably higher plasma cortisol concentration in most clinical cond-tions, the lack of specificity may be of minor importance, but must be considered in evaluating increased plasma cortisol values. The mean plasma cortisol concentration in normal subjects (20.3 mug/100 ml) was equal to or slightly higher than found by other methods. The kit was found appropriate for routine clinical application, due to the easy, rapid and inespensive performance with an acceptable recovery and reliability, but it cannot entirely replace more specific methods.