Objective To describe the characteristics, management, and outcomes of women undergoing invasive therapies for primary postpartum haemorrhage (PPH).Design A population-based observational study.Setting All 106 maternity units of six French regions. Methods Prospective identification of women with PPH managed with invasive therapies, including uterine suture, pelvic vessel ligation, arterial embolisation, and hysterectomy.Main outcome measures Rate of use and failure rate of invasive therapies, with 95% confidence intervals (95% CIs).Results An invasive therapy was used in 296 of 6660 women with PPH (4.4%, 95% CI 4.0-5.0), and in 0.2% of deliveries (95% CI 0.18-0.23). A hysterectomy was performed in 72/6660 women with PPH (1.1%, 95% CI 0.8-1.4%), and in 0.05% of deliveries (95% CI 0.04-0.06). A conservative invasive therapy was used in 262 women, including 183 (70%) who underwent arterial embolisation and 79 (30%) who had conservative surgery as the first-line therapy. Embolisation was more frequently used after vaginal than caesarean delivery, and when arterial embolisation was available on site. The failure rate of conservative invasive therapies was 41/262 (15.6%, 95% CI 11.5-20.6) overall, and was higher after surgical than after embolisation procedures, in particular for vaginal deliveries.Conclusions Both maternal mortality as a result of obstetric haemorrhage and the rate of invasive therapies used for PPH are high in France. These findings suggest flaws in the initial management of PPH and/or the inadequate use of invasive procedures.Keywords Embolisation, hysterectomy, maternal mortality, pelvic vessel ligation, postpartum haemorrhage, uterine compression sutures.Tweetable abstract Maternal mortality as a result of haemorrhage and the rate of invasive therapies used for PPH are high in France.