With limited research on the mortality of crack users, this 5-year follow-up study examined a cohort of 131 crack-dependent patients admitted to the detoxification ward of a public hospital in Brazil between 1992 and 1994. The study sample was predominately young, male, White, of low educational attainment, and unemployed, with 60 percent of the sample using crack for more than 1 year. Participants were interviewed using a structured questionnaire. Seventeen variables were tested as predictors of mortality using Cox’s proportional hazards regression, such as demographic data, level of schooling, drug use variables, criminal history, treatment history and pre-admission psychosocial assessment. Three variables were identified as predictors of mortality: history of intravenous drug use, unemployment at index admission, and premature discharge during index admission. Additional findings indicate homicide as the main cause of death (56 percent) and that the use of crack substantially increases the risk of a violent death. Study limitations and recommendations for future research were presented. References