Serum concentrations of prostatic acid phosphatase, prostate-specific antigen, neopterin, osteocalcin, tissue polypeptide antigen, and CA-50 were measured before onset of treatment in 86 patients suffering from prostatic carcinoma. In an attempt to identify patients with a poor prognosis, the data were related to patient survival after 3 years. When the standard reference values, which are based on studies on healthy subjects, were used in the study on the deceased, the diagnostic sensitivity was acceptable, whereas the diagnostic specificity was low. A better relation to prognosis was obtained if higher discrimination values were employed. The increased specificity could be obtained with little loss of sensitivity. It is suggested that the discrimination value should be chosen after careful consideration of the expected performance of the test in differentiating between defined groups of patients. To indicate short-term prognosis in prostatic cancer, a higher than normally recommended discrimination value for the marker should be selected.