The acquired immune deficiency syndrome (AIDS) is a disease that is transmitted almost entirely through behavioral factors. In the absence of a cure or vaccine, the modification of AIDS-risk behavior presents a unique challenge to behavioral scientists and should be taken as a clear imperative by behavior analysts. This paper discusses the currently dominant social-cognitive theories (the health belief model, the theory of reasoned action, and self-efficacy theory) that have been widely used to predict and understand AIDS-risk behavior. Although these theories have generated a voluminous literature on the cognitive, attitudinal, and demographic correlates of AIDS-risk behavior, they have not resulted in specific intervention strategies to influence risky behavior, most likely because they fail to specify manipulable variables. As an alternative to social-cognitive theories, this paper evaluates the usefulness of a behavior-analytic approach to stem the spread of HIV infection. It examines some of the philosophical differences underlying cognitive and behavioral approaches that are embedded in mechanistic versus functional contextualistic principles. It explores the theoretical and practical implications of adopting either predicting and explaining behavior or predicting and influencing behavior as the goals of science. To illustrate the value of adopting the goal of prediction and influence, behavior-analytic research on the social context of risky sexual behavior in adolescents is described. The paper argues that in order to alter the future course of the AIDS epidemic, the behavioral sciences must move beyond describing cognitive and attitudinal correlates of risky behavior and focus on the social context of the behavior of individuals. In addition, population-wide changes in AIDS-risk behavior can be accomplished only if research focuses on how to influence larger social systems, including the media, school systems, and community organizations.