A finding in the 1975-76 evaluation of several California programs designed to improve the academic achievement of young children indicated that in schools whose students scored at or below the twentieth percentile on the Entry Level Test in 1973-74, third grade reading test scores had declined relative to their predicted scores over the three-year period from 1973-74 to 1975-76. Although the aggregate finding for that group of schools indicated a decline in scores relative to prediction, individual schools could be identified that had patterns of scores that increased over the period in question, as well as schools in which scores declined. This study was designed to investigate the circumstances associated with this decline or improvement in third-grade reading achievement. Seven "increaser" schools and seven "decreaser" schools were identified, and a total of 45 third-grade classrooms were observed. Reading test scores from the California State Assessment of 1976 and 1977 and attendance data were obtained for the year. Major findings included the following: (1) higher classroom gains were associated with lower teacher-student ratios; (2) highly controlled classroom environments using systematic instruction and much positive reinforcement contributed to greater reading achievement scores; (3) small group instruction was more efficient and effective in increasing reading achievement than was individualized instruction; and (4) great classroom diversity was observed in terms of instructional processes and achievement gains. Appendices include the classroom observation instrument, and operational definitions for the observation instrument. (Author/DB)