The study conducted domestic-violence screening for 2,616 women as they were admitted into a hospital or health care clinic for any kind of treatment. An attempt was made to interview all women who screened positive and approximately 30 percent of the women who screened negative. Approximately 66 percent of the abused women who were interviewed were reinterviewed at least once during the following 12 months. The homicide sample included all of the 87 intimate-partner homicides in 1995 or 1996 that had a woman victim or offender age 18 or older. Up to three people who were familiar with the relationship (friends, family, or the woman herself when she was the perpetrator) were interviewed. Questionnaires used in the interviews contained questions on such topics as household composition, mental and physical health, substance use, firearm availability, social support network, power and control, harassment or stalking, and help-seeking. Women who had experienced violence in the past year developed a "calendar history" of every violent incident and other important events that had happened in the year. Findings show that violence in the past predicted violence in the future. The majority of women who were killed or who killed their partners had experienced violence from their partners in the previous year. The features of the incident were also found to be significant in increasing the risk for death and severe injury. Risk was heightened when the violence involved the threat or use of a weapon and having been choked or strangled; when it followed shortly after a previous incident; and when there had been an increase in the frequency of violence. For 27 percent of the 143 clinic/hospital women who experienced only 1 incident in the previous year, that incident was life threatening. For these women, important risk factors were the partner's controlling behavior, especially jealousy; the partner's drug use; and the partner's violence outside the home. In 45 percent of the homicides in which a man killed a woman, an immediate precipitating factor was the woman leaving or trying to end the relationship. Women who killed their partners had experienced more severe and increasing violence in the previous year than other victims, and they had fewer resources, such as employment or a high school education. They also had a more traditional relationship with their abuser (married, had children, or were in a long-term relationship). More than 30 percent of the clinic/hospital women who had experienced severe or life-threatening violence in the previous year had not sought any kind of formal help. Implications of these findings are drawn for researchers and for practitioners. 7 references