Background Amniotic fluid embolism (AFE) remains one of the principal reported causes of direct maternal mortality in high-income countries. However, obtaining robust information about the condition is challenging because of its rarity and its difficulty to diagnose. This study aimed to pool data from multiple countries in order to describe risk factors, management, and outcomes of AFE and to explore the impact on the findings of considering United Kingdom, international, and United States AFE case definitions. Methods and findings A population-based cohort and nested case-control study was conducted using the International Network of Obstetric Survey Systems (INOSS). Secondary data on women with AFE (n = 99-218, depending on case definition) collected prospectively in population-based studies conducted in Australia, France, the Netherlands, Slovakia, and the UK were pooled along with secondary data on a sample of control women (n = 4,938) collected in Australia and the UK. Risk factors for AFE were investigated by comparing the women with AFE in Australia and the UK with the control women identified in these countries using logistic regression. Factors associated with poor maternal outcomes (fatality and composite of fatality or permanent neurological injury) amongst women with AFE from each of the countries were investigated using logistic regression or Wilcoxon rank-sum test. The estimated incidence of AFE ranged from 0.8-1.8 per 100,000 maternities, and the proportion of women with AFE who died or had permanent neurological injury ranged from 30%-41%, depending on the case definition. However, applying different case definitions did not materially alter findings regarding risk factors for AFE and factors associated with poor maternal outcomes amongst women with AFE. Using the most liberal case definition (UK) and adjusting for the severity of presentation when appropriate, women who died were more likely than those who survived to present with cardiac arrest (89% versus 40%, adjusted odds ratio [aOR] 10.58, 95% confidence interval [CI] 3.93-28.48, p < 0.001) and less likely to have a source of concentrated fibrinogen (40% versus 56%, aOR 0.44, 95% CI 0.21-0.92, p = 0.029) or platelets given (24% versus 49%, aOR 0.23, 95% CI 0.10-0.52, p < 0.001). They also had a lower dose of tranexamic acid (median dose 0.7 g versus 2 g, p = 0.035) and were less likely to have had an obstetrician and/or anaesthetist present at the time of the AFE (61% versus 75%, aOR 0.38, 95% CI 0.16-0.90, p = 0.027). Limitations of the study include limited statistical power to examine factors associated with poor maternal outcome and the potential for residual confounding or confounding by indication. Conclusions The findings of our study suggest that when an AFE is suspected, initial supportive obstetric care is important, but having an obstetrician and/or anaesthetist present at the time of the AFE event and use of interventions to correct coagulopathy, including the administration of an adequate dose of tranexamic acid, may be important to improve maternal outcome. Future research should focus on early detection of the coagulation deficiencies seen in AFE alongside the role of tranexamic acid and other coagulopathy management strategies.