Several inconsistent research results regarding the effect of Problem-Based Learning (PBL) in enhancing Mathematical Abilities (MA) of primary school students have been reported widely by some researchers. To summarize, estimate, and evaluate the effect of PBL on MA of primary school students and investigate the study characteristics that provoked the heterogeneous effect size, a systematic review and meta-analysis was employed. Hedge's equation was employed to measure effect size using Comprehensive Meta-Analysis (CMA) software as an analysis tool. The synthesis of 16 relevant primary studies revealed that the overall implementation of PBL had a high positive effect (g = 1,050) significantly compared to conventional learning on MA of primary school students based on a random effect model. Moreover, the characteristics of the sampling technique, publication index, and MA significantly caused the heterogeneity of the effect size of PBL towards students' MA. Thus, these results suggest that primary school mathematics teachers should select PBL as one of the best solutions to enhance students' MA and consider the characteristics of students MA that will be improved. Besides, researchers should select random sampling as a sampling technique to select the sample and the literature indexed by Scopus as references to support the quality of research.