This study develops and tests an instrument to assess the fidelity of the intended program, i.e., experimental treatment in the evaluation of a preschool program. During the school year (1972-73) CEMREL (Central Midwestern Educational Lab) investigated the consequences of different levels of training on implementation of the Demonstration and Research Center for Early Education (DARCEE) program,. Part of this investigation involved three separate ratings of the pilot test classrooms with the assessment scale. These ratings were given at the beginning, middle, and end of the school year. Classrooms with maximum training scored on the average approximately 10 per cent higher on each of the essentials than did the classes with materials only. With comparison classes, however, that consistency was lacking. On the essentials of physical setting, unit use, and parent involvement the comparison classrooms actually scored higher than the DARCEE group with the maximum training, on two other essentials (reinforcement and behavior management and attitude development) and on student involvement they scored higher than DARCEE classrooms with the minimum training, whereas on the essentials of skill development, organization and use of time, grouping, teacher roles and responsibilities, and teacher preparation these classes scored lower than both DARCEE classroom treatments. (Author)