A retrospective review of 64 rectocele repairs done over a four-year period was performed. The most common indication for repair was constipation. Thirty-five patients were repaired transanally, and 29 were repaired transvaginally. The overall morbidity was 34 percent, and the overall mortality was 0 percent. The most common complication was urinary retention in 12.5 percent. There was no difference in complications between techniques. Of 46 patients contacted for follow-up, 25 (54 percent) still complained of constipation, 17 (34 percent) had partial incontinence, 8 (17 percent) noted persistent rectal pain, 15 (32 percent) mentioned occasional rectal bleeding, and 10 (22 percent) complained of vaginal tightness or sexual dysfunction. Thirty-seven (80 percent) patients stated that they had improved after surgery. Except for persistent rectal pain, there was no difference in results between transanal and transvaginal repairs. Those undergoing transvaginal repair had a much greater problem with pain. Our relatively poor results may be due to an unselective approach to rectocele repair. The presence of both constipation and a rectocele does not imply an association, and a complete anorectal physiologic examination should precede repair. There is no functional difference between transvaginal and transanal rectocele repair.