BACKGROUND:Severe postpartum haemorrhage (PPH) is a serious clinical problem that is increasing in incidence. AIM:To identify risk factors for severe PPH. MATERIALS AND METHODS:Population-based retrospective cohort study of all women who gave birth in Victoria in 2009-2013 using the validated Victorian Perinatal Data Collection. Three multivariable logistic regression models estimated the adjusted risk of severe PPH. Adjusted odds ratios (aOR) and their 95% confidence intervals are reported. The primary outcome was severe PPH (estimated blood loss of ≥1500 mL). RESULTS:Severe PPH occurred in 1.4% of all births (n = 5122). Maternal characteristics significantly associated with severe PPH included: multiple pregnancy; older maternal age; overweight/obesity; first births. Other risk factors included placental complications, macrosomia, instrumental vaginal birth, third and fourth degree perineal lacerations, in-labour caesarean section, birth at a gestation other than 37-41 weeks, duration of labour 12 to <24 h, and use of oxytocin infusions in labour. Planned pre-labour caesarean section was associated with reduced odds of severe PPH. Severe PPH also occurred in 0.7% (n = 104) of women with none of the identified risk factors. CONCLUSIONS:Numerous risk factors for severe PPH are identified but some cases are not modifiable or predictable. Limiting use of oxytocin infusions in labour to cases with clear indications, and strategies to prevent severe perineal lacerations would prevent some severe PPHs. Close surveillance of all women in the hours immediately following birth is crucial to detect and manage excessive blood loss and reduce severe PPH and associated morbidity.