Teachers who do not package appropriate learning models and media for science learning so that students find it challenging to learn. This study aimed to develop a PBL-based video. This type of research is developed using the ADDIE model. The research subjects were one content expert, one learning design expert, and one learning media expert. The test subjects were 3rd-grade students for individual trials, nine small group trials, and 27 people for field trials. Methods of data collection using questionnaires or questionnaires. The data collection instrument used a rating scale instrument. Data analysis techniques are qualitative descriptive statistical analysis, quantitative descriptive statistical analysis, and inferential statistics. The results of the analysis, namely the validation of media experts, obtained a percentage of 93.33% (very good), material experts 98.68% (very good), and design experts 89.70% (very good). The individual test results were 93.33% (very good), the small group trial was 95% (very good), and the field test was 94.35%. Based on the results of the analysis shows that the significance level is less than 0.05 (p<0.05). Thus, it can be seen that there is a significant difference in student learning outcomes for science content before using PBL-based learning videos. It is concluded that PBL-based learning videos improve student learning outcomes.