We analyzed complexed and free prostate-specific antigen (PSA), the free/total PSA and complexed/free PSA ratios, acid phosphatase, and prostatic phosphatase in serum from 36 patients with prostatic carcinoma and from 48 non-neoplastic control patients (20 with prostatitis and 28 with benign prostatic hyperplasia). Receiver-operating characteristic plots showed that serum PSA was the most efficient variable, singly used, in discriminating neoplastic from non-neoplastic patients. At a cut-off value of 10.0 ng/ml, serum PSA had a diagnostic sensitivity of 87% and a diagnostic specificity of 83%. In particular, three patients with prostatic carcinoma and twenty non-neoplastic controls had serum PSA levels of between 4 and 10 ng/ml. The subsequent analysis of the serum free/total PSA ratio, in this subgroup, using a cut-off level of 15%, allowed us to classify correctly all prostatic cancer cases and 18/20 non-neoplastic diseases. We next analyzed PSA mRNA in circulating cells using an improved reverse-transcriptase polymerase chain reaction dot blot procedure, from six patients with prostatic carcinoma with distant metastases, and in seventeen with localized cancer. The analysis had a high sensitivity (up to dilutions 1:10(6) of total RNA from prostatic cancer cells vs total RNA from normal blood cells). The analysis revealed circulating micrometastatic cells in 3/6 (50%) cases of metastatic cancer and in 4/17 cases of localized cancer. To conclude, serum total PSA combined with the free/total PSA ratio is a very efficient algorithm in discriminating neoplastic from non-neoplastic prostatic diseases, while other mRNA species must be analyzed, in addition to PSA mRNA, in circulating cells to increase the efficiency in detecting metastatic prostatic cancer.