Much research has argued for the existence of a marriage benefit to men, women, and children. Although the commonly suggested current response to a husband's violence has been for the couple to separate, traditionally women were often told to "stay for the children." This paper uses the Project on Human Development in Chicago Neighborhoods longitudinal data to examine the relationships among marital stability, husband violence, and children's behavior problems. Among married mothers who reported husband violence at time 1 (N = 414), the negative relationship between child behavior problems and remaining in the marriage dissipated when selection bias and husband violence recidivism were controlled.