We reviewed our frozen section experience with 310 pelvic lymphadenectomy specimens during the last 5 years. A total of 40 patients (12.9 per cent) had positive lymph nodes on permanent section. In 6 of these patients the lymph nodes were involved grossly and in 34 there were only microscopic metastases. Intraoperative assessment of lymph node involvement classified correctly 299 patients (96.5 per cent of the total number). Whereas previous studies have demonstrated a failure on frozen section to detect all but a few microscopic metastases, we were able by frozen section to identify metastases in 23 of 34 patients (67.6 per cent) with grossly uninvolved lymph nodes. Of the positive frozen sections 16 were in patients with unilateral metastases only, and in 13 of these cases frozen section identified the only positive node present. The average diameter of the metastases found on frozen section was 2.4 mm. In 11 of the 34 patients frozen section did not disclose any of the metastases present on permanent sections (average 1.4 mm.) (false negative rate 3.5 per cent of all patients, 27.5 per cent of those with positive nodes and 32.3 per cent of those with microscopic involvement of lymph nodes only). Ten patients had unilateral metastases and 1 had bilateral involvement. The ability to identify the majority of microscopic metastases, given their adverse effect on prognosis, supports the usefulness of routine frozen section on grossly uninvolved pelvic lymph nodes as a staging procedure before radical prostatectomy.