In this paper we study the effects of abusive behavior on the labor force andmarital status decisions of women. Using a unique Canadian data set on domesticviolence, we estimate the effects of abuse on the marital history as well ascurrent employment using a sequential, multi-state model. In our model, spousalabuse affects labor supply through decreases in utility from leisure as well asthrough reductions in productivity at work and hence the market wage. Inaddition, abuse is treated as an initially unobserved spousal characteristicthat plays a role in the divorce decision, which in turn influences laborsupply. Our analysis reveals three main findings. First, the effects ofdomestic abuse on employment differ across marital histories. Employment isdecreasing in the presence of abuse in current and past marriages for marriedand divorced women, respectively, consistent with a health effect on the wage.In contrast, remarried women are more likely to work if abused in the current,but not the past, marriage. Second, domestic abuse is a dominant factor in thedivorce decision, which in turn is a major determinant of employment. Finally,standard economic information such as age and education plays a minor role inthe divorce decision relative to the abuse-related information.