The authors present the results of a community-wide infant mortality review, describe implications for the delivery of maternal and child health services, and discuss the value of such reviews in addressing local public health concerns. The review included an analysis of birth and death certificates and medical record data; maternal interviews; review of cases and development of recommendations by provider panels; and convening of community groups to develop strategies to improve the health and health care of women and infants. The review focused on 287 infant deaths during 1990-1993. More than half of all neonatal deaths were attributable to "previable" or "borderline viable" births. Sexually transmitted infections were the most frequently identified underlying risk, and smoking was the most frequently identified prenatal risk. Homelessness, physical and sexual abuse, and alcohol use were at least twice as likely among women whose babies died than among a high risk comparison group. Panelists identified fragmented health care over the course of women's reproductive lives as a predominant theme. The authors conclude that: (a) The focus of maternal and child health care should shift to a model of women's health care that addresses the chronicity of social and clinical risks. (b) Infant mortality reviews are a valuable tool for community education, systems review, and policy development and can be applied to other public health issues with local significance. (c) Expectations about the review process's ability to produce conclusions about causality or recommendations narrowly geared to reducing infant mortality rates need to be reframed. (d) The model will be strengthened by greater participation of families affected by infant death.