The aim of this research is to identify the implicit cognitive models of AIDS held by individuals in two high risk groups--clients seeking HIV screening and prison inmates. Identification of these models is important because earlier research suggests that such cognitive representations of illness control a variety of health related behaviors, including compliance with health instructions. Although AIDS education has increased how much people know about the disease, research to date suggests only limited effective behavior change to reduce risky sexual and drug-related behaviors. Such research suggests some undefined cognitive or affective component which mediates between instructional messages and health behaviors. It is suggested that effective interventions must address the specific cognitions that control risky sexual and drug-related behaviors. The practical value of the research, then, is to identify those specific targets and to learn if there are significant population differences specific targets and to learn if there are significant population differences in models of AIDS that would require different educational targets. On a theoretical level, previous research with other illnesses has produced mixed results, suggesting both a general model for a variety of illnesses and distinctive models for particular illnesses. No research to date has examined personal cognitive representations of AIDS. It is proposed that because of their cultural and psychological differences, the prison and clinic populations may hold different cognitive models of the disease. Responses to a questionnaire (derived from earlier research) by large samples of clinic patients and prison inmates will be subjected to exploratory and confirmatory factor analyses. For Study 1 (clinic patients), demographic data and behavioral information regarding sexual and drug behavior will be correlated with factors of the model. Models will be compared by subjecting each model to confirmatory procedures based on data from the other sample. In later analyses, data will be regrouped so that potential sex and race effects may be explored.