OBJECTIVE: To compare blood loss with spontaneous delivery and manual removal of the placenta during caesarean section. DESIGN: A randomised controlled trial. SETTING: Four university hospitals between September 1999 and June 2002. POPULATION: A total of 472 women delivering by caesarean section at term were randomised to spontaneous placental delivery (n= 235) or manual removal (n= 237). METHODS: The allocation was made by opening the next available of a series of sealed opaque envelopes and derived from a computer-generated list of numbers. MAIN OUTCOME MEASURES: Significant blood loss, defined as either a drop in haemoglobin of greater than 2.5 g/dL, or the need for blood transfusion. RESULTS: The mean interval between delivery of the newborn and the placenta was longer in the spontaneous delivery group (3.4 vs 1.9 minutes), but the mean duration of the operation was similar. Significant blood loss occurred in 30 women (13%) in the spontaneous delivery group and 49 women (21%) in the manual removal one (RR 0.62; 95% CI 0.41-0.94). Post-operative fever affected 6 and 5 cases, respectively, and antibiotics were used in 14 and 12 cases, respectively. CONCLUSIONS: Allowing spontaneous delivery of the placenta reduces significant blood loss without increasing operating time.