Objective: The goal of the study was to evaluate the surgical procedures used to manage recurrent stress urinary incontinence in a tertiary referral center, to compare the procedures with respect to efficacy and failure rates, and to identify risk factors for failure. Study Design: The health records of patients who underwent surgical treatment of recurrent stress urinary incontinence performed by the senior author (H.P.D.) between 1984 and 1995 were reviewed. The objective cure rate was evaluated by means of urodynamic studies and physical assessment, and the subjective cure rate was determined by means of historical report. In light of our selection criteria, the time to failure, the number of previous anti-incontinence procedures needed to give the best cure rate, and the risk factors for failure of the 3 predominantly used surgical techniques were determined. The statistical methods used were the χ2 test with 95% confidence interval, the Cox proportional hazard model with logistic regression, and survival analysis. Results: One hundred ninety-eight female patients were surgically treated for recurrent stress urinary incontinence between January 1, 1984, and December 31, 1995. Four surgical techniques were used: (1) the combined abdominovaginal (2-team) polypropylene (Marlex; Phillips Petroleum Company, Bartlesville, Okla) sling (group 1, n = 70), (2) the modified urethral Marlex sling (group 2, n = 68), (3) Burch retropubic colpourethropexy (group 3, n = 49), and (4) suburethral Marlex sling (group 4, n = 11). The study population consisted of 118 patients. The rest of the original 198, including all the patients in group 4, were excluded. Objective and subjective cure rates of 69% and 89%, 66% and 96%, and 69% and 88% were calculated for groups 1, 2, and 3, respectively. By 6 years after the operation 100% of the failures in groups 1 and 2 had occurred, and 88% of the group 3 failure occurred within 2 years after the operation. Cure rates of 77%, 73%, and 38% (P = .320) were achieved with the 2-team sling procedure (group 1) after 1, 2, and 3 previous anti-incontinence operations, respectively, whereas cure rates of 81%, 25%, and 0% (P = .001) were obtained with the Burch procedure (group 3) after 1, 2, and 3 previous anti-incontinence operations, respectively. This indicates that the Burch procedure should be avoided after >1 previous operation, whereas the 2-team sling can be used after ≥3 previous anti-incontinence procedures. Statistical significance could not be determined for group 2 because that procedure was not used to treat any patient with 3 previous operations. Age was a marginal risk factor for failure in group 1. No statistically significant risk factors were identified for group 2. The number of previous anti-incontinence procedures was the major risk factor for failure in group 3 when age, parity, gravidity, weight, hormone replacement therapy, number of previous anti-incontinence procedures, and urethral closure pressure were covariables. Conclusion: According to our data, both sling procedures and Burch retropubic colpourethropexy can be used to surgically manage recurrent stress urinary incontinence with selection criteria such as those that we used. In our opinion comparative prospective studies of different surgical techniques with similar selection criteria, long-term follow-up of ≥10 years, and the inclusion of urodynamic studies may be the most ethical way to determine the right operations for recurrent stress urinary incontinence. (Am J Obstet Gynecol 1999;181:1296-309.)