Evidence-based research on the effect of Realistic Mathematical Education (RME) has been conducted. However, whether differences in the measured variables alter the studies' effect sizes have not been explored. To fill this gap, it is necessary to conduct a meta-analysis study that can summarize the evidence for the effect of RME and analyze how the differences in the measured variables are associated with the effect size (ES) of the study. The research sample is an individual study of the effect of RME on various mathematical abilities identified from the ERIC database, the Scopus database, and Google Scholar. Based on the inclusion requirements, 54 independent samples from 38 individual studies were included in the analysis with a total of 6140 students included. The estimation method uses a random-effect model, and the Comprehensive Meta-Analysis (CMA) software is used as a data processing tool. The results of the analysis showed that the ES of the study was 0.97. This shows that overall, the use of RME has a significant effect on students' mathematical abilities. The moderator analysis results explain that the differences in the measured variables moderate the implementation of RME. These findings contribute to the implementation of RME in classrooms and the further development of RME by considering the categories of abilities being measured.