BACKGROUND: Physiotherapy involving pelvic-floor muscle training is advocated as first-line treatment for stress urinary incontinence; midurethral-sling surgery is generally recommended when physiotherapy is unsuccessful. Data are lacking from randomized trials comparing these two options as initial therapy. METHODS: We performed a multicenter, randomized trial to compare physiotherapy and midurethral-sling surgery in women with stress urinary incontinence. Crossover between groups was allowed. The primary outcome was subjective improvement, measured by means of the Patient Global Impression of Improvement at 12 months. RESULTS: We randomly assigned 230 women to the surgery group and 230 women to the physiotherapy group. A total of 49.0% of women in the physiotherapy group and 11.2% of women in the surgery group crossed over to the alternative treatment. In an intention-to-treat analysis, subjective improvement was reported by 90.8% of women in the surgery group and 64.4% of women in the physiotherapy group (absolute difference, 26.4 percentage points; 95% confidence interval [CI], 18.1 to 34.5). The rates of subjective cure were 85.2% in the surgery group and 53.4% in the physiotherapy group (absolute difference, 31.8 percentage points; 95% CI, 22.6 to 40.3); rates of objective cure were 76.5% and 58.8%, respectively (absolute difference, 17.8 percentage points; 95% CI, 7.9 to 27.3). A post hoc per-protocol analysis showed that women who crossed over to the surgery group had outcomes similar to those of women initially assigned to surgery and that both these groups had outcomes superior to those of women who did not cross over to surgery. CONCLUSIONS: For women with stress urinary incontinence, initial midurethral-sling surgery, as compared with initial physiotherapy, results in higher rates of subjective improvement and subjective and objective cure at 1 year.