The main obstacle to good empirical outcome research is the sociopolitical atmosphere stimulated by the issue of child maltreatment. The resulting pressure for action has prevented the unhurried attitude and slow accumulation of knowledge necessary to the scientific approach to problem solution. The result has been unclear definitions of target populations, interventions, and outcomes. Careful attention to definitions, sampling, and matching of control groups would help address these problems. Another major problem is inherent in the nature of child abuse itself. Child abuse is a private event that occurs with low frequency, requiring lengthy followup periods for outcome evaluation. However, child maltreaters tend to be poor, less well-educated, highly stressed, socially isolated, and mobile. Their chaotic lifestyles hinder both treatment participation and research efforts. For example, they may be unable to read or understand written questionnaires. In addition, the violence that characterizes their lives can endanger both researchers and subjects. The resulting methodological problems include small sample sizes, selection bias, limited scope and sources of dependent measures, poor treatment specifications, limitations in experimental design, and limited practical significance of results. Increased funding, networking of research projects, and greater methodological care are needed to address these problems. 18 references.