The purposes of this study were: (1) to investigate the relationship between self-regulation, motivational belief, and task performance; (2) to compare the differences of self-regulation, motivational belief, and task performance respectively between students with different item-selection strategies; (3) to compare the differences of the difficulty level of the first item between students with different item-selection strategies and to analyze the relationship between task performance and the difficulty level of the first chosen item; and (4) to evaluate the effects of self-regulation training course. The subjects were 90 sixth-grade students (42 boys and 48 girls) drawn from Ku-Ting Elementary School in Taipei City. Each subject was given Self-Regulation Inventory, Motivational Belief Inventory, and Arithemtic Computation Task (A, B). Forty-Five subjects of the entire sample received the self-regulation trining course. The data were analyzed with product-monent correlation, canonical correlation, one-way ANOVA, mixed design two-way ANOVA, and Bowker´s test of symmetry. The findings were as follows: (1) Self-regulation had significant correlation with motivational belief, and the latter accounted for 43% of the former´s variance. The two factors-intrinsic value and achievement motivation-of motivational belief had significant positive correlation with task performance; but, self-regulation was not related to task performance. (2) Between different item-selection strategies, there were differences on self-evaluation and self-revision of the process of self-regulation; there was also significant difference on achievement motivation which was one of the three factors of motivational belief. In addition, there was a difference on task performance between students with different item-selection strategies. The sutdents who used failure-tolerant stategy got higher scores than those who used the other strategies on self-evaluation, self-revision, and achievement motivation. (3) There were differences on the difficulty level of the first selected item bwteen students with different item-selection strategies. The student who used failure-tolerant strategy began with more difficult items than those who used other item-selection strategies. (4) with the self-regulation training course, the experimental group obtained higher scores in self-regulation and motivational belief than those of the control group; but no difference was found between their task performance. These results support the viewpoint of Pintrich & DeGroot (1990) and Zimmerman (1986) that motivation factor had influences on the use of self-regulation. The findings of this study that subjects´ abilities of self-regulation can be improved with self-regulation training are also in congruence with the result of DeJong´s research (1990).