Data on mode implementation were obtained through participant observation, staff questionnaires, youth interviews, and monthly data abstraction. The measurement of program implementation addressed the four program design dimensions: a multiple phase program, theoretical principles, structural elements, and the actual intervention strategies. Findings indicate that program implementation can be measured by adhering to a range of validity criteria for implementation. There was no single element or configuration of elements that determined implementation. Implementation occurred when a threshold of elements was established. The combination of elements varied in the strongest innovations, but these sites exceeded the level of implementation required to establish an acceptable program environment. The translation of the reintegration emphasis in the well-implemented programs was similar: job and school preparation, reliance on transitional placements between institutions and community, and reconstruction of family and neighborhood ties. These programs were notable for clear rules and contingencies, strong modeling of prosocial behaviors, problemsolving, advocacy in the community, and differential approaches to risk and treatment. 7 tables, 45 references, survey instrument.