The use of self-correction strategy helps students to detect their own mistakes and self-correct them, thereby students self-regulate their own learning processes. It is still a strategy that is rarely researched in mathematics learning, particularly at the primary level. This study attempts to fill in the gaps by determining the effectiveness self-correction strategy on primary school students' mathematics achievement. It also aims to determine the mediating roles of problem-solving self-efficacy on the relationships between the two variables. To assess the effects of self-correction strategy empirically, a Pretest-Posttest Nonequivalent Control Group design was employed. The study involved 62 primary school students. The experimental group (n=32) applied self-correction strategy during 12 mathematics lessons while the control group (n=30) did not. Achievement of both groups was measured by the mathematics test. Students' problem-solving self-efficacy was measured by the Problem-Solving Self-Efficacy Questionnaire. Results of Analysis of Covariance (ANCOVA) found that self-correction strategy was effective in improving the experimental group's mathematics achievement [F(1, 59) = 11.00, p < 0.05, eta square = 0.157]. The results of regression analysis showed that self-efficacy was a significant mediator between the use of self-correction strategy and their mathematics achievement [F(1, 61) = 27.60, p = 0.046, [beta] = -0.201]. Implications of the findings from the educational and research standpoints were discussed and suggestions for future research were provided in this paper.