Two predictions related to educational performance are examined in this experiment: (1) Competitive treatments increase classroom performance and interest on a power test; and (2) game-like competition is as effective as reward competition in increasing classroom performance and interest on a power test. 1,035 fifth and sixth graders from 36 classes in four Wisconsin school systems participated in the research and were randomly assigned to either a control group (competition with reward treatment), or an experimental group (competition in a game setting treatment). Forms 4A and 4B of the mathematical subtest of School and College Ability Tests were used to measure treatment effects. The competitive treatments had no significant effect on performance but significantly increased the childrens' interest in the test. There was no difference between the two significant treatments; competition with a game was as effective as competition with a reward for both performance and interest measures. The speculation is made that the relative magnitude of the performance and interest effect is directly related to the nature of the task. That is, competition increases performance more in a speed task than in a power task; and competition increases interest more in a power task than in a speed task. (Author/RSM)