As a significant tool to integrate information technology and education, dynamic mathematical software (DMS) has been widely concerned in recent years. However, how to better apply it to instruction practice deserves further exploration. Thus, we adopted the meta-analysis method to analyze the DMS-based experiments published between 2000 and 2020. A three-level meta-analysis of data from 107 studies involving 10,507 participants and 138 effect sizes revealed a moderate effect size (d = 0.632, 95% CI = [0.550, 0.713]). Moreover, moderator analyses showed that: (1) cultural background had significant moderating effects; (2) students performed better on near-transfer tests than far-transfer tests; (3) DMS used by students independently had better effects; (4) intervention duration had significant moderating effects; (5) some of the above significant moderating effects were unique after controlling for others. Overall, our findings suggest that DMS has positive effects on students' performance and teachers should be meticulous in designing their teaching plans.