This study aims to measure the effect size of experimental research that focuses on the application of the blended learning model in science learning and measures students' scientific competence. The method in this study used a meta-analysis technique with a quantitative approach. We found 11 articles that matched the criteria with a range of 2015-2021 publication years from SINTA accredited national journals and reputable international journals. The data tabulation sheet was used as a research instrument containing criteria in the form of experimental and control groups by containing information on the number of samples, average, standard deviation, education level, country, type of control, and students' science skills. The data were analyzed using the openMEE application and it obtained a summary effect size of 0.861 so that it can be interpreted as being in the high category and it is assumed that blended learning is effective in improving students' science skills in the form of conceptual understanding, science process skills, critical thinking, and others. Blended learning can be recommended as learning that has the potential to be implemented in the current digital era and can be a reference for further researchers to develop blended learning models in science learning.