In order to inform policy and programs, research on child abuse must be not only methodologically rigorous, but also practically feasible. However, practical problems make child abuse research difficult to conduct. Definitions of abuse must be explicit and different types of abuse must be assessed separately. Study samples should be as representative as possible of the universe of abused children. Comparison groups of nonabused children are necessary to determine whether the effects of abuse are independent of other factors known to influence child development. These groups should be similar in life experiences and circumstances to the abused groups. Considerable attention must be given to methods for increasing response rates. Maintaining study samples over time requires diligence. These problems are by no means the only ones likely to arise, but they are common to many studies. Access to officially reported cases requires the cooperation of protective service agencies. Thus, research objectives must fit overall agency goals and findings must be useful for policy and program development. Investigators must provide cooperating agencies with reports of study findings. Although research will not solve the problem of child abuse, high caliber research is needed to build a knowledge base about the causes, correlates, and consequences of abuse that will ultimately lead to solutions. (ABL)