This study aims to analyze the effect of problem-based learning in physics learning that is reviewed as overall, school level, research results or response variables, subject matter, and the media used. This research method uses meta-analysis. The meta-analysis research stage is to determine the research topic, determine data selection criteria, search for data, classify data information, then analyze the data and making conclusion. The data collection technique uses secondary data from previous research of 23 scientific publication articles. The data analysis technique uses the effect size. The results of this meta-analysis concluded that the implementation of problem-based learning had a great effect on learning physics with an average effect size of 0.524. The application of problem-based learning in physics learning had a great effect when applied at the tertiary level, increasing creative thinking skills, using light as the subject matter, and assisted by virtual learning media in the form of a video-based laboratory.