In a comparative study, seven different methods for the determination of cortisol in human plasma were evaluated, using routine patient samples. Four of these methods used radioactive steroids (125I- or 3H-labelled) and in three no radioactivity was needed. For the statistical evaluation a direct 3H-radioimmunoassay was arbitrarily taken as the independent variable. It was found that all other methods correlated well with this assay. However, the simplest method, the fluorimetric, cannot be recommended mainly because of its non-specific fluorescence and troublesome interference from some widely-used drugs. Of the methods evaluated a radioimmunoassay is recommended. For laboratories having no equipment for measurements of radioactivity, the more elaborate and time-consuming fluorometric method of Clark may represent a good alternative.