This study intended to clarify developmental changes of causal schemata in ability and effort attributions of academic attainment, and to examine individual differences in the development of causal schemata by comparing underachievers and overachievers. In the first study, we made a pair comparison questionnaire designated as Inference Test consisting of three parts. First, the subjects were asked to infer academic attainments of two hypothetical persons described on the level of their abilities and efforts, and to indicate the one who would gain higher academic attainments. Second, the subjects were requested to answer which person would have higher ability after being given information on the level of academic attainment and effort of two hypothetical persons. Third, the subjects imagined two hypothetical persons reading information on the level of their academic attainment and ability, and were asked to infer which person would have made more effort. The subjects were 26 third graders, 27 fourth graders, 41 fifth graders and 40 sixth graders in an elementary school, and 66 university students as adults. Our analysis of the data led to the following results. 1.Regarding with inference of academic attainment from ability and effort information, most of the third graders could use the so-called graded schema when they were given information on two persons of similar ability but different on effort. That is to say, they would believe the person making more effort would gain higher academic attainment. But two third of the third graders could not use the graded schema when shown information on two persons different on ability but equal on effort. 2.In the case of comparing two hypothetical persons ability according to the given academic attainment and effort information, most of the fourth graders could use the magnitude covariation schema, which was confirmed by the positive relation between the given degrees of success and the inferred degrees of ability(or effort). Most of the third and fourth graders thought that, when two hypothetical persons performed equally well, the one who exerted more effort was also more capable. This could mean that the so-called hallo schema tended to be used. About 70 percent of all sixth graders, however, possessed the inverse compensation schema leading to the inference that, if an effect remained invariant, a change in the magnitude of one cause was accompanied by a compensating change in the magnitude of a second cause. 3.As for inferring effort from academic attainment and ability information, the magnitude-covariation schema was used by more than 80 percent of the third graders, with an inverse compensation schema emerging among the fourth and fifth graders. In general, the subjects could use more complex causal schemata when inferring effort than when inferring ability. In the second study, the subjects were underachievers and overachievers in a junior high school;the study aimed at comparing these two groups from a developmental standpoint of an inverse compensation schema. Another questionnaire like Inference Test was made asking the subjects the degree of one cause(i.e., ability or effort)about a hypothetical person having information on the level of academic attainment and other cause(i.e., effort or ability respectively). The data were analyzed separately in success and failure situations, classified on the basis of the level of academic attainment. In comparing overachievers and underachievers in relation to the use of the inverse compensation schema, the former could use the schema significantly more frequently than the latter when ability was inferred. This suggests that causal schemata in overachievers was more highly developed than in underachievers. When effort was inferred, however, the two groups were not different.