In an ideal world, effective programs would be disseminated on a large scale, but many obstacles block program diffusion. To date, little guidance has been provided by program designers on how to implement the program in different sites with different personnel and client populations. This article discusses the debate between the "fidelity camp" and the "reinvention camp" and suggests an approach that reconciles the two positions: fidelity should be maintained to the program's mechanism of operation; reinvention is permitted in implementation if the causal mechanism is preserved. Such an approach is essential for effective dissemination of preventive interventions that are successful in controlled trials.