This research uses a quantitative approach with experimental methods. The experimental design applied was one group pretest-posttest design. The population and sample in the research were students in class VA SDN 64/I Muara Bulian, totaling 27 students. The sampling technique used in this research was Probability Sampling with Cluster Random Sampling type. The data collection technique is a problem solving ability test. Validation of the assessment instrument uses validity tests and reliability tests. Data analysis techniques through normality and hypothesis testing activities. The research results show that the PMRI approach has quite an influence on the problem solving abilities of fifth grade elementary school students. This is proven by the average initial ability value, namely 47.34, and the final ability after being given the PBL model treatment, namely 68.60. According to N-Gain calculations, the average increase from pretest to posttest is 0.43 in the medium category. After carrying out the Wilcoxon test, the sig value was also obtained. < 0.05, namely 0.000 < 0.05, meaning Ha is accepted and Ho is rejected. Thus, it can be stated that the PMRI approach has quite an influence on the problem solving abilities of fifth grade elementary school students.