21st century learning has had a huge impact on students and teachers. Students must be guided to be able to think critically to overcome various kinds of problems that occur. Furthermore, teachers as education play an important role in honing problem solving skills in their students. The problem solving learning model is a learning model that directs students to solve problems in the learning process. In addition, the problem solving learning model is not fully capable of being applied by science teachers. Although, in the 2013 curriculum in Indonesia a teacher is required to be able to apply the problem solving model in conducting science learning. Not only that, the limited experience of teachers in applying variations of these learning models is still the main obstacle in learning science. So, teachers in Indonesia must have extensive experience in adopting problem solving learning models. This study aims to determine the effectiveness of the problem solving learning model in learning science in Indonesia. This research is a kind of meta-analysis research. The research sample came from analyzing national and international journals published from 2010-2022. The selection of data used as samples was carried out very strictly and thoroughly. Searching for this research sample through Google Scholar, Eric, Hindawi, Sage, Springer, Proquest, IEEE, DOAJ, and Wiley. The sample selection technique is a purposive sampling technique. In this technique, the data that is used as a sample must have a relationship with the research variable. In searching for sample keywords, namely problem solving learning models in learning science (Biology, Chemistry and Physics) in various schools in Indonesia. Data analysis is descriptive statistical analysis with the OpenMEE application. Data analysis was performed by calculating effect size, standard deviation (SD), average value (mean), and N-gain. The results of this study concluded that the application of the problem solving learning model was very effective in learning science in Indonesia. This learning model is able to increase student learning outcomes with an average score of 86 students in the experimental class and 65.5 in the control class, the effect size value is 1.39 in the high category and N-Gain is 0.55.