Teacher professional development (PD) is essential to continuously improve teaching skills, to adapt to diverse student needs, and to promote equity and inclusion. Only a few studies to date have synthesized how PD programs improve teachers' content knowledge and instructional quality, as well as students' academic performance. In this meta-analysis, we aim to evaluate the impact of PD programs on science teachers and their students. We calculate a total of 514 effect sizes using Hedges' g from 66 studies published between 2010 and 2022. The bias-corrected standardized mean difference (Hedges' g) is within-subject and between-subjects design. The overall effect size is 0.772, indicating a substantial effect size on PD effectiveness (s.e. = 0.063, p < 0.001, 95% CI [0.647, 0.897]). We observe considerable heterogeneity of effect sizes, moderated by PD dosage hours, duration, and active teaching. The findings indicate that relatively short PD periods--less than 48 h--and durations under 3 months may yield the most effective science PD for educators. As an alternative, slightly longer engagement exceeding 72 h, coupled with sustained support over 6 months, has proven to be the second most effective option for PD. Furthermore, an active learning approach within PD programs (g[bar] = 0.794, s.e. = 0.066, 95% CI [0.656, 0.926]) has emerged as a pivotal influence on PD effectiveness. This study provides insights into education research and policy to understand PD research and to ensure how PD can be designed and implemented to improve student performance.