We conducted a cohort study assessing risk factors for developing urinary incontinence following childbirth, and a pilot randomized controlled trial of a physiotherapist-led intervention to reduce incidence of incontinence. A total of 723 women were recruited to the cohort study, of which 234 entered the nested trial and were randomized to intensive training in pelvic floor exercises or standard information. At 6 months post-partum, 45% of women reported some incontinence problems. A pre-existing incontinence problem was the best predictor of future incontinence (odds ratio 4.49, 95% confidence interval (CI) 3.09-6.53). Chronic constipation (1.86, 1.03-3.34) and episiotomy in at least one delivery (1.96, 1.25-3.07) were also independent risk factors, while an epidural or spinal (0.62, 0.42-0.92) was protective. The intervention as designed did not help in preventing future incontinence (relative risk 1.28, 95% CI 0.98-1.67), but this may be due to the failure to persuade the women to return for the classes. Any intervention aimed at promoting postnatal pelvic floor exercises should be limited to women who have already been experiencing incontinence problems.