The study sample came from three sources: Edna Mahon Correctional Institution for Women in Clinton, New Jersey (105 women); two suburban New Jersey police departments (50 women); and two battered women's shelters in New Jersey (55 women). Women in the prison sample had been convicted of killing their partners and were serving sentences ranging from 3 years to life. The majority of incarcerated women had been convicted of second degree murder or manslaughter. Most battered women who killed their abusive partners were much more likely than the nonhomicidal group to have dropped out of high school, to have an erratic work history of one or two unskilled jobs, to be cohabitating with their partners, to have drug problems, to have attempted suicide by overdosing on drugs, to have received emergency medical treatment for battering-related injuries, and to have had access to the batterer's guns. In contrast to the homicidal group, battered women in the community sample were much less likely to be alcoholics or drug addicts, have experienced alcohol-related blackouts, have received psychiatric treatment, have attempted suicide, and have access to a gun. Findings support the idea that once a battered woman receives a death threat and fails to "drown her sorrows" in alcohol or drugs and to commit suicide, she is likely to try to kill her batterer. 23 references and 3 tables