The long history of gender bias, patriarchal culture, and social science misinformation militate against sex-fair analyses and suggest that researchers make their values explicit. Development of an ethical code may prevent the continuation of poorly conceived research in the area of family violence. Toward such a code, it is suggested that research should do no harm to those who are studied who are still affected by family violence. Research should be context-specific and should be evaluated for experimenter bias. Sample populations should be carefully selected, and the limits of generalization should be made specific. Researchers should be responsible for the societal uses of data collected and protect the sample from intrusive legal scrutiny. Programs also should be designed to minimize stresses on researchers. While researchers should be supportive they should avoid dual and sexually exploitive relationships with research subjects. Student research should be supervised by knowledgeable faculty or outside consultants, and research problems to be studied should be chosen in consultation with service providers. 41 references.