A meta-analysis was conducted of 29 comparative studies located primarily through computer searches of Comprehensive Dissertation Abstracts and ERIC. Studies chosen took place in actual elementary classrooms, reported measured outcomes for both computer-based education and control classes, and were free from methodological flaws. A total of 21 variables were used to describe aspects of the studies' experimental treatments, methdology, settings, and publication history. Outcomes were also coded, with student learning the most commonly measured outcome. Results showed that computer-based education has generally had positive effects on the achievement of elementary school pupils. These effects have been different, however, for programs of offline computer managed instruction (CMI) and for interactive computer assisted instruction (CAI). In 25 studies, CAI drill and practice and tutorial programs raised student achievement scores by 0.48 standard deviations (or from the 50th to the 68th percentile). In four studies, CMI programs raised student achievement scores by only 0.07 standard deviations. It is noted that study features were not significantly related to study outcomes. Forty-five references are listed. (LMM)