Several authors have suggested that the purpose of evaluating programs in their dissemination phase is to show that the conditions necessary to produce the expected outcomes are implemented. The need for outcome evaluation should thus decrease when a program goes from the demonstration to the dissemination phase. Disputing such a proposition, this paper argues that: 1) the evaluation agenda should be broad and open in all phases of program development; 2) the need for outcome assessment should be present during the entire cycle of any program's life, and 3) there is a need to develop creative methodological approaches to address issues relevent to the evaluation of dissemination programs. Finally, eight research strategies are presented for addressing two such issues: the relationship between the researcher and the program and the equivocality of the causal inference present during the dissemination phase.