Transrectal ultrasound (TRUS) was performed preoperatively in 35 patients with rectal carcinoma and the results were compared to histologic findings. In the same group, postoperative studies were performed in 22 patients; in women, transvaginal ultrasound (TVUS) was added to the transrectal study. According to Duke's classification modified by Astler-Coller, in relation to the "T" parameter, TRUS correctly staged 33 of 35 neoplasms (accuracy, 94.3%); one was overstaged and one was understaged. In detection of lymph node involvement, accuracy was 74% (sensitivity 69%, specificity 73.9%). Recurrent local tumors, histologically confirmed, developed in two of 22 postoperative patients who had undergone curative anterior resection. This study demonstrates that TRUS is an accurate method in preoperative staging of rectal carcinoma. In the prospective study, the role of follow-up TRUS and TVUS in detection of local recurrences is evaluated.