This article reviews the literature that compares the instructional effectiveness of games to conventional classroom instruction. Studies dealing with empirical research rather than teachers'judgments are reviewed. Published reviews of research in English dating from 1963 to 1984 were examined and the literature was searched for studies from 1984 to 1991. Of the 67 studies considered over a period of 28 years, 38 show no difference between games and conventional instruction; 22 favor games; 5 favor games, but their controls are questionable; and 3 favor conventional instruction. Results for social sciences, math, language arts, logic physics, biology, retention over time, and interest are examined. Math is the subject area with the greatest percentage of results favoring games, but only eight studies have adequate controls. Thirty-three out of 46 social science games/simulations show no difference between games/simulations and classroom instruction. The authors conclude that subject matter areas where very specific content can be targeted are more likely to show beneficial effects for gaming.