Objective: To evaluate the relation between midline episiotomy and postpartum anal incontinence. Design: Retrospective cohort study with three study arms and six months of follow up. Setting: University teaching hospital. Participants: Primiparous women who vaginally delivered a live full term, singleton baby between 1 August 1996 and 8 February 1997: 209 who received an episiotomy; 206 who did not receive an episiotomy but experienced a second, third, or fourth degree spontaneous perineal laceration; and 211 who experienced either no laceration or a first degree perineal laceration. Main outcome measures: Self reported faecal and natus incontinence at three and six months postpartum. Results: Women who had episiotomies had a higher risk of faecal incontinence at three (odds ratio 5.5, 95% confidence interval 1.8 to 16.2) and six (3.7, 0.9 to 15.6) months postpartum compared with women with an intact perineum. Compared with women with a spontaneous laceration, episiotomy tripled the risk of faecal incontinence at three months (95% confidence interval 1.3 to 7.9) and six months (0.7 to 11.2) postpartum, and doubled the risk of flatus incontinence at three months (1.3 to 3.4) and six months (1.2 to 3.7) postpartum. A non-extending episiotomy (that is, second degree surgical incision) tripled the risk of faecal incontinence (1.1 to 9.0) and nearly doubled the risk of flatus incontinence (1.0 to 3.0) at three months postpartum compared with women who had a second degree spontaneous tear. The effect of episiotomy was independent of maternal age,infant birth weight, duration of second stage of labour, use of obstetric instrumentation during delivery,and complications of labour. Conclusions: Midline episiotomy is not effective in protecting the perineum and sphincters during childbirth and may impair anal continence.