OBJECTIVE: To determine the changes in maternal mortality rates over the 19-year period from 1985 to 2003 at Point G National Hospital, Bamako, Mali. METHODS: Data on all pregnant women admitted from January 1, 1985 to December 31, 2003 were collected from all hospital services. Records were entered into a database, and maternal mortality rates and cause-specific fatality rates were analyzed. RESULTS: Significant declines in the fatality rates due to uterine rupture (odds ratio [OR] 0.086; 95% confidence interval [CI], 0.011-0.70) and postpartum infection (OR 0.22; 95% CI, 0.085-0.55) were noted when the period from 1985-1987 was compared with 2001-2003. Overall, there was a significant reduction in the odds of maternal death in the period between 1985-1987 and 2001-2003 (OR 0.529; 95% CI, 0.341-0.821), when adjustments were made for cesarean delivery, complications, and patient age. CONCLUSIONS: Although the crude maternal mortality rate remained high, shifts in the patient population that led to more patients with complications being admitted masked improvements in the odds of death for obstetric patients.