We reviewed all cases of massive primary postpartum haemorrhage greater than 1000 mL over a six month period in 1999 to establish the incidence, identify aetiological factors and implement change. Fifty-four cases (1.7%) were identified. We classified four as 'near-miss' maternal mortality. Over 60% were delivered by caesarean section. Seventy-six percent were due to uterine atony, 9% due to genital tract trauma and 15% were associated with significant antepartum haemorrhage from placenta praevia or abruption. No obvious labour or delivery risk factors were identified but deviation from hospital guidelines was common. Following revision of the guidelines, dissemination to staff and use of practice drills, we repeated the study on a prospective basis over the same time period in 2002. There was a significant reduction in the incidence of massive postpartum haemorrhage to 0.45%, and 100% adherence to the guidelines which resulted in a significant reduction in maternal morbidity. We believe that this approach can be replicated in other units.