Model programs for chronic mental patients may be viewed from four perspectives: evaluation of individual programs, commonalities in successful programs, generalizability and reproducibility of specific programs, and relevance of model programs to problems of service delivery in mental health systems. Although successful model programs share certain common structural elements, such programs cannot be readily reproduced or generalized. Having limited value for the problems of service delivery in mental health systems, model programs are best seen as experimental efforts, not as solutions. Strategies for translating model-derived knowledge into systems-related action are needed.