The occurrence of complications during pregnancy depends less on the degree of human development than differences in the way complications in pregnancy are detected and managed. It is the quick diagnosis and correct management that really contribute to the enormous differences in maternal mortality ratios between countries and regions. Understanding of the determinants of maternal mortality may be improved by studying cases of severe maternal morbidity. In this paper, various approaches to the concept of severe maternal morbidity and near-misses are discussed, and the relationship between these and maternal deaths. Although no consensus has been reached on a strict definition of near-miss or severe maternal morbidity, we show that the definitions used may be tailored to support diverse objectives, including monitoring progress, epidemiological surveillance and auditing of health care. We conclude that the versatility of the concept, the greater frequency of cases available for study and the possibility of interviewing the survivors of severe complications all support the value of studying severe maternal morbidity to help guide local efforts to reduce maternal mortality. Although this may almost be a reality in developed countries, it continues to represent an important and difficult challenge to overcome in places where its benefits would be most evident.