The design of the flipped classroom (FC) as a pedagogical approach is becoming increasingly prevalent in mathematics education. However, although many studies have been conducted on whether FCs are more effective in improving mathematics achievement compared to traditional lectures, the results are not consistent. To address this issue, this meta-analysis examined the overall effect of flipped mathematics classrooms on student achievement, in addition to a set of moderating variables. Our analysis included thirty-seven experimental studies with forty effect sizes, comparing flipped and traditional classrooms in K-12 and tertiary mathematics education. The meta-analysis revealed an overall statistically significant FC effect of g = 0.402, indicating a marginally moderate effect size. The effect sizes were also significantly moderated by education level and mathematics content area. No evidence of publication bias was found in the data set. We concluded by discussing employment of the FC model in the contexts of mathematics education and offering suggestions for future research.