OBJECTIVE The reverse transcriptase-polymerase chain reaction (RT-PCR)-prostate-specific antigen (PSA) assay to detect presumed occult micrometastatic prostate cancer has been controversial, and this molecular staging has been thought to be clinically useful by some groups but not others. METHODS We used a sensitive nested RT-PCR assay with specific primers derived from the PSA sequence and a very stringent two-step PCR protocol with denaturing temperature of 94 degrees C annealing and extension temperature of 68 degrees C. This method enabled us to detect PSA-expressing LNCaP prostate cancer (PC) cells as low as one cell of 10 million lymphocytes (1/10(7)). Ninety-six patients with PC were studied, including 85 before radical prostatectomy (RP), and 22 controls, including healthy men and women and men with benign prostatic hyperplasia. RESULTS In 85 patients undergoing RP, a minimum of two independent RT-PCR-PSA assays detected circulating prostate cells preoperatively in 27 patients (31.8%). Of 12 patients with locally advanced or advanced stage cancer, RT-PCR-PSA was positive in 5 (41.7%); of the 22 controls, no patient was RT-PCR-PSA positive. In 10 randomly selected cases, the RT-PCR product was confirmed as PSA by DNA sequencing. Of the 27 patients undergoing RP who were RT-PCR positive, 11 (40.7%) had non-organ-confined disease (pT3a or greater), and of the 58 patients who were RT-PCR negative, 32 (55.2%) had non-organ-confined disease. Patients with RT-PCR positive results also had lower margin positivity (9 of 27, 33.3%) than did patients with RT-PCR negative results (21 of 58, 36.2%). Finally, at a mean follow-up of 25.7 months, 5 (18.5%) of 27 RT-PCR positive patients had recurrence (PSA) compared with 14 (24.1%) of 58 RT-PCR negative patients. CONCLUSIONS On the basis of this blinded study, RT-PCR for PSA-expressing cells in 85 patients before RP is not related to clinical stage, age, race, grade, Gleason sum, serum PSA or prostatic acid phosphatase, tumor volume, or tumor multifocality. RT-PCR positivity did not predict pathologic stage or early PSA recurrence. A standardized RT-PCR assay needs to be developed to account for interlaboratory discrepancies.