Research to develop and ensure diffusion of smoking prevention programs must (a) be based on an appreciation of the social, psychological, and biological determinants at each stage in the onset process, (b) disentangle major interactions between program content, participant, provider, and setting factors as they determine impact, and (c) ensure both that diffusion is based on empirically grounded principles and that the process is monitored and its effectiveness evaluated. Sufficient evidence supports the tentative conclusion that social influence curricula can be efficacious--at least with some youth. However, we lack key information for diffusion, in particular concerning provider and setting factors. Thus, a cautious advance to diffusion research is recommended, noting that there is much we do not know, and that the public health need for applications must be balanced with continuing research to clarify for whom and under what circumstances current curricula work. At the same time, there should be strong continuing research to improve current interventions, especially for high risk populations.