OBJECTIVE: To describe a method for identification, classification and assessment of maternal deaths in Denmark and to identify substandard care. DESIGN: Register study and case audit based on data from the Registers of the Danish Medical Health Board, death certificates and hospital records. SETTING: Denmark 2002-2006. POPULATION: Women who died during a pregnancy or within 42 days after a pregnancy. METHODS: Maternal deaths were identified by notification from maternity wards and data from the Danish National Board of Health. A national audit committee assessed hospital records of direct and indirect deaths. MAIN OUTCOME MEASURES: Maternal mortality ratio, causes of death and suboptimal care. RESULTS: In the study period, 26 women died during pregnancy or within 42 days from direct or indirect causes, leading to a maternal mortality ratio of 8.0/100,000 live births. Causes of death were cardiac disease, thromboembolism, hypertensive disorders of pregnancy, Streptococcus A infections, suicide, amniotic fluid embolism, cerebrovascular hemorrhage, asthma and diabetes. CONCLUSION: Our method proved valid and can be used for future research. Causes of death could be identified and learning points from the assessments could form the basis of focused education and guidelines. Future complementary 'near miss' studies and cooperation with other countries with comparable health systems are expected to improve the benefits of the enquiries, contributing to improved management of life-threatening conditions in pregnancy and childbirth.