Plasma carcinoembryonic antigen concentration, serum phosphohexose isomerase activity and the erythrocyte sedimentation rate were determined in patients with cancer, patients being treated for various benign conditions and healthy young persons to make an intercomparison between the ability of these assays to distinguish patients who have cancer from those who do not and from normal persons. The upper boundaries of the assays were defined by the one-sided, distribution-free, upper tolerance limits of the reference populations. Only by means of the assay for carcinoembryonic antigen was it possible to distinguish patients who had cancer from those who did not. Twenty-two percent of patients with cancer could be identified by this assay with a false-positive error rate of < 5%.