To determine the methodological quality of studies evaluating surgery for stress incontinence, the effectiveness of different procedures and the frequency of complications associated with each procedure. Eleven randomized controlled trials, 20 non-randomized trials/prospective cohort studies and 45 retrospective cohort studies were reviewed systematically. The methodological quality of the 31 prospective studies was generally poor. The considerable variation in inclusion criteria, surgical management and assessment of outcome precluded any statistical meta-analysis. Evidence as to the effectiveness of surgery for stress incontinence is weak; therefore, any conclusions are speculative. It appears that colposuspension may be more effective and the effect more long-lasting than that following anterior colporrhaphy and needle suspension. There is little information on the value of sling procedures. Comparisons of different ways of performing each procedure show no significant differences in outcome but this may reflect the methodological weaknesses of the studies. Valid and reliable data on the frequency of complications following surgery are lacking so the safety of the procedures is unclear. Repeat operations to correct stress incontinence are less successful than first procedures but this finding may be subject to confounding. There is an urgent need for some large, rigorous, prospective studies of high quality. Until such studies have been completed, recommendations as to the best clinical practice cannot be based on scientific evidence. Studies need to define cases according to widely accepted criteria, including standard measures of the severity of stress incontinence, and surgical terminology for the procedures performed needs to be standardized and outcomes need to be clearly defined, valid and reliable, not confined to short-term assessment and include patients' views along with the surgeon's assessments.