This study examined the objectivity, trainability, and reliability of the SRI observational system (developed at the Stanford Research Institute) as it was modified for use in the "A Study of Schooling" research project. Four instruments in the observation system were tested. The physical environment inventory recorded the architectural arrangement of the classroom, seating and grouping patterns, and materials. The daily summary provided an overview of available space and materials as well as of student and teacher decision making processes. The classroom snapshot provided information about what each person in the classroom was doing at a given point in time and the nature of classroom activities. The five minute interaction was a continuous accounting of how time was spent in the classroom, focusing on the interactive process between teacher and students. By placing two observers in 357 elementary and secondary school classrooms, inter-observer reliability was analyzed. The results indicated that observers could be trained to use the SRI Observational System and that the instruments provided reliable data for analysis of classroom observations. Unreliability occurred mostly for behaviors which were rarely observed. Specific similarities and problems found in the use of each instrument are discussed. Appendices present data on correlations between observers. (CJ)