This review examines the current state of knowledge on the effectiveness of HIV prevention interventions with respect to sexual behavior among the population of drug users. The review focuses specifically on the relation between intervention methods found to be effective and the underlying theory. Electronic searches were conducted and supplemented by publications gathered through other channels. The studies were reviewed for (1) study design, (2) evident use of theory in intervention development, (3) clear targeting of determinants, (4) description of the study or studies, and (5) evaluation of the behavioral goals and targeted determinants. For each study, a description is given of: (1) the size and nature of the sample; (2) the retention rate; (3) the study design; (4) the nature of the intervention programs, including theoretical methods, practical strategies and theoretical background; (4) the measures of variability. The results show that a limited number of interventions were effective in changing sexual risk behavior among drug users. More successful programs featured several of the following elements: use of multiple theories and methods, inclusion of peers and rehearsal of skills. Moreover, the community-level interventions showed the importance of sustainability. The most successful intervention methods were modeling, skill building and social support enhancement. These methods are generally derived from the Social-Cognitive Theory or the Diffusion of Innovations Theory. Future HIV/AIDS interventions should build on the strengths discussed. Evaluations of interventions should be designed to facilitate comparison, using standardized and specific behavioral outcomes as well as standardized and preferably long-term follow-up levels, and should also evaluate the impact of programs at a psychosocial level to examine whether or not the theoretical methods on which a program was based were actually effective in changing the psychosocial factors targeted and why.