Described here is a research project being undertaken to address two points: the relationship of the fidelity of program replication to program outcomes at user sites; and the extent to which fidelity can occur in programs with more than a few replicates. The findings will have bearing on the usefulness of the Research, Development, and Diffusion (RD & D) Model, which has been criticized on the grounds that it does not take into account the modification or "reinvention" of programs that occurs when they are replicated at other sites. In the current study, eight programs are being examined--four in education and four in the criminal justice field. Fidelity of replication is to be determined using Hall and Loucks's methods of measuring the number and importance of program components implemented. Data have already been collected by telephone interview with program developers and implementers and will also be gathered during site visitations. The relationship between fidelity and program outcomes will be analyzed using a multiple regression approach. Although the research is at an early stage of data collection, it can already be concluded that specificity in detailing program components is critical in conducting both evaluation and implementation research. (Author/JM)