Introduction - The National Confidential Survey on Maternal Mortality (ENCMM) aims to compile data on all maternal deaths in order to draw general conclusions about the quality of care. This report covers the periods 2001–2003 and 2004–2006. Methods - The definitions used are those of the World Health Organization (WHO). The statistical sources are civil registry data (births and deaths processed by INSEE), medical causes of death (CépiDc), and obstetric data from the ENCMM (Inserm - Unit 953). A simplified data collection procedure was used for deaths occurring between 2004 and 2006. New estimates of maternal mortality have been calculated (adjusted rates per 100,000 live births), by age, mother’s nationality, and region. The characteristics of the deaths (location, gestational age, autopsy, detailed obstetric causes, “preventability”) are presented as a percentage of deaths. Results - The adjusted maternal mortality rate is estimated at between 8 and 12 deaths per 100,000 live births. The risk of maternal death is three times higher among women aged 35–39 than among those aged 20–24, and remains higher among women of foreign nationality. More than three-quarters of deaths occur in resuscitation or intensive care units. Direct obstetric causes are by far the most common (73%), due to hemorrhage (25%) and amniotic fluid embolism (11%). Hemorrhages, infections, and anesthetic complications are mostly “preventable.” None of the downward trends in frequency are statistically significant over this period. Conclusion—Despite generally satisfactory outcomes, care does not benefit the entire population equally: certain subgroups of women may require preventive policies or more appropriate care. (R.A.)