For the third year, six Massachusetts communities incorporated computer assisted instruction (CAI) into their regular Title I programs, for students in grades 1 through 9. It was hoped that CAI would produce reading and mathematics gains over and above those which are characteristic of more conventional Title I instructional approaches. A pretest, posttest quasi-experimental research design was employed to determine the program's effectiveness. Title I students were assigned to treatment (CAI) and control (or comparison) groups. All participating students received 30 minutes per day of Title I instruction in reading and/or mathematics; students in the CAI group spent 10 of those 30 minutes interacting with the computer system. Data were analyzed across and within grade levels. Students in the CAI groups were found to demonstrate significantly superior performance to non-CAI students in reading, and superior (but not statistically significant) performance in mathematics. The uniform results obtained were attributed to the high degree of design compliance in the past year, as compared to previous years of the program's operation. (GC)