BACKGROUND: Recognition of the synergy between health and poverty is now apparent in the development strategies of many low-income countries, and markers are needed to monitor progress towards poverty-relevant goals. Maternal mortality has been proposed as a possible candidate but evidence is lacking on the link with poverty at the level of individuals. We introduce a new approach to exploring the relation--the familial technique. METHODS: We used data from 11 household surveys in ten developing countries to create percentage distributions of women according to their poverty-related characteristics and survival status (alive, non-maternal death, maternal death). These women were identified as the sisters of the adult female respondents in the surveys, and were assigned the same poverty status as their respondent sibling. FINDINGS: The analysis showed significant associations, across a diverse set of countries, between women's poverty status (proxied by educational level, source of water, and type of toilet and floor) and survival. These associations indicated a gradient within and across the survival categories. With increasing poverty, the proportion of women dying of non-maternal causes generally increased, and the proportion dying of maternal causes increased consistently. Further analysis reported here for one of the countries--Indonesia, revealed that about 32-34% of the maternal deaths occurred among women from the poorest quintile of the population. The risk of maternal death in this country was around 3-4 times greater in the poorest than the richest group. INTERPRETATION: This new method makes efficient use of existing survey data to explore the relation between maternal mortality and poverty, and has wider potential for examining the poor-rich gap.