This paper describes some activities that author has designed using Relating, Experiencing, Applying, Cooperating, and Transferring (REACT) learning on students' problem solving mathematics abilities. This study used a quasi-experimental study with a nonequivalent posttest only control group design. The subject in this study are 20 elementary students. Data were collected using a questionnaire and a test and this analyzed using descriptive statistical analysis techniques and inferential statistics (t-test). The results showed that the positive response of students using REACT learning to improve mathematics students’ understanding of problem solving abilities, namely 14,25 is average yes and 5.75 is average no. Furthermore, the results of hypothesis analysis through the t test, namely t count of 30.822 which means t count> t table at or 30.822> 1.734 with a significant level of 5%. It means this learning give a significant effect on students' problem solving mathematics abilities.