The field of social innovation policy can presently be seen as divided into two opposing camps: pro-fidelity and pro-adaptation. The former conceptualizes innovations as consisting of a number of relatively well specified components, and argue that rigorously developed, evaluated programs should be implemented with close correspondence to validated models. The latter argues that differing organizational contexts and practitioner needs demand on-site modification. To provide empirical evidence about the pro-fidelity position, seven innovative social programs developed and disseminated nationwide by various public sector organizations (schools, city agencies, prisons, courts) using federal funds were studied. Methods for measuring program fidelity and effectiveness across sites were developed. Results supported the two pro-fidelity assumptions. Four of the seven programs exceeded the acceptable level of fidelity, while the remaining three were in the acceptable range. The second assumption was supported by a significant correlation between fidelity and effectiveness. However, despite the overall support, there was considerable cross-program variation. (Author/JAC)