Three types of drill were proposed to test the effectiveness of individualization on drill practice. The division problems to be practiced in the experiment were sequenced according to difficulty level of problems (called problem's level) by predicted value estimated from equation 1. The subjects were divided into one of five ability levels by pretest. In a drill of branching type the drill broblems of the 1st day were matched in learner's ability level. During subsequent practice days problem's level assigned to him depended on his last result (FIG. 1). In a drill of mixedtype an ability level decided by pretest was appliedto learner throughout drill practice period. Half of the drill problems in this type were composed of problems which were coincident with learner's ability level. The remaining were formed from other problem's level (TABLE 2). In a fixed type learner was given the problems made from all problem's difficulty level (TABLE 3). After four drill practice days the posttest was performed. The main results were as fallows: 1. In analysis of variance of TABLE 5, the difference of total score was unsignificant among three drill practice types. However, as can be seen in a significant drill practice tests interaction, there's obviously difference in gains from pretest to posttest: A fixed type is best with proportion correct of gains of 0.163 (number of correct answers, 4.88), a mixed type is next with 0.125 (number of correct answers, 3.64), and a branching type is lowest at 0.070 (number of correct answers, 2.10). 2. Proportion correct was higher as the learner's ability level increased and probleml's difficulty level decreased in all drill practice types (TABLE 6, 7, 8). 3. Gains from pretest to posttest depended on learner's ability level in both fixed and mixed type. Learner with higher ability obtained a little gains (ability 4 and 5). Learner with moderate ability showed the greatest gains (ability level 2 and 3). But learner of low ability reversely indicated the decrease especially in a fixed type from pretest to posttest. Such interaction was not found in a branching type. These results were distinctly opposed to the prediction that practice effect was higher as drill practice was individualized to learner's ability. Then, it was suggested that individualization of drill practice resulted in restricting learner's potentiality because of missing the opportunity to challenge harder problems.