Computational thinking is believed to be beneficial for Science, Technology, Engineering, and Mathematics (STEM) learning as it is closely related to many other skills required by STEM disciplines. There has been an increasing interest in integrating computational thinking into STEM and many studies have been conducted to examine the effects of this intervention. This meta-analysis examined the effects of computational thinking integration in STEM on students’ STEM learning performance in the K-12 education context. Following systematic procedures, we identified 20 publications with 21 studies meeting the inclusion and exclusion criteria from a range of academic databases. We extracted effect sizes on student learning outcomes in one-group pretest-posttest designs. We also examined a range of moderating variables in the models, including student levels, STEM disciplines, intervention durations, alignment with content standards (e.g., CSTA/NGSS), types of intervention (e.g., simulation), and the use of unplugged/plugged activities. Overall, we found a statistically significant large effect size ( g = 0. 85 [95% CI of 0.57–1.14]; p < .001), indicating a large overall effect of computational thinking integration on STEM learning outcomes. The effect sizes were significantly moderated by intervention durations. We provide a discussion of the findings and present implications for future research and practice.