Since the passage of the Pennsylvania Protection from Abuse Act in 1976, many States have enacted legislation to provide civil restraining orders for battered women. These orders, which offer a civil court alternative to criminal sanctions, are court-issued temporary or permanent orders which direct an assailant to refrain from further abusive conduct. However, interviews with 270 recipients of restraining orders suggest that the orders are generally ineffective in reducing the rate of abuse or violence. Analysis indicates that postproject abuse and violence are unaffected by the presence of a restraining order. Nearly three victims in five were abused within 4 months regardless of whether they had obtained a restraining order. While the restraining orders were ineffective in stopping physical violence, they were effective in reducing abuse for women with less serious histories of family violence or where the assailant was less violent in general. Measures to improve restraining order mechanisms should more clearly codify abuse and violence, improve access for those not married or cohabitating, streamline procedures and shorten waiting periods, address a full range of child-related concerns, strengthen sanctions, and mandate official responses to violations. Additionally, comprehensive legislation is needed to coordinate civil and criminal remedies. Four tables and 23 references are included. (Author abstract modified)