This retrospective study of 100 cases of primary carcinoma of the vagina treated at Magee-Women's Hospital of the University of Pittsburgh from 1962 to 1992 analyzes disease and treatment-related prognostic factors and treatment results. Fifty percent of patients had previous hysterectomies prior to the diagnosis of vaginal cancer. The tumor was limited to one-third of the vaginal canal in 64 patients, but involved more than one-third of the canal in 36 patients. Treatment consisted of surgery in 40 patients, radiation therapy in 47 patients, and surgery plus radiation therapy in 13 patients. With a median follow-up of 11.2 years, the 5-year actuarial disease-free survival (DFS) was 67% for Stage I (23 patients), 53% for Stage II (58 patients), 0% for Stage III (9 patients), and 15% for Stage IV (10 patients). Using univariate and multivariate analysis, treatment with surgery, disease limited to one-third of the vaginal canal, and FIGO Stage I and II disease were significantly favorable prognostic factors for DFS. Treatment with surgery was superior to radiation therapy alone in Stage II patients (P = 0.00004). There appeared to be a selection bias in Stage II patients, as the more unfavorable cases were treated with radiation therapy. Although radiation therapy will often be the treatment of choice, excellent results and limited morbidity accompany selected cases surgically resected.