OBJECTIVE: To compare the outcome, 12 months after surgery, of three principal categories of procedure (colposuspension, needle suspension and anterior colporrhaphy) used for treating stress incontinence. METHODS: A non-randomised trial design comparing 221 colposuspensions, 54 needle suspensions and 130 anterior colporrhaphies performed by 49 surgeons in 18 hospitals in the North Thames health region in 1993-1994. Four outcomes were considered: complications, severity of stress incontinence, social impact of incontinence, and activities of daily living score. Results were adjusted, using logistic regression, for 13 case-mix variables. RESULTS: Significant differences existed between the three procedures in the characteristics of the patients. The cure rate varied by procedure (colposuspension 34% dry; needle suspensions 13%; anterior colporrhaphy 19%). Two thirds of women, however, reported an improvement (colposuspension 75%; needle suspension 68%; anterior colporrhaphy 55%). After adjusting for confounders, colposuspension was significantly more likely to result in an improvement than anterior colporrhaphy (odds ratio 2.2). While this was reflected in improvements in the social life of the women, the difference between procedures did not reach statistical significance (p = 0.05). CONCLUSION: These results suggest surgery for stress incontinence in typical settings is not as good as reported in textbooks. This is mostly because this study was based on women's own reports (and not surgeons' reports) and partly because of the unselected nature of the patients. The information on outcomes given to patients should be reviewed and a large, pragmatic randomised trial including patient-assessed outcomes is needed.