The purpose of the current study was to explore college women's stay/ leave decisions in abusive relationships using a prospective methodology. Participants (N = 323) completed surveys at the beginning and end of a 10-week academic quarter for course credit. A path analysis suggested that the model-which included investment model variables (i.e., relationship commitment, investment, satisfaction, and quality of alternatives), childhood abuse, psychological distress, avoidance coping, and self-esteem-was a good fit to the data and predicted abused women's leaving behaviors over the interim. The implications of these findings for future research, theory, and clinical work are discussed.