This research is motivated by the low mathematical communication skills of students at school which causes students to be less involved in learning or students are rarely given the opportunity to ask questions or express opinions so that students' mathematical communication skills have not been able to be conveyed. This study uses the Indonesian Realistic Mathematics Education (PMRI) approach to see the differences between PMRI learning and conventional learning models for increasing mathematical communication skills in terms of student gender. The type of experiment used in this research is Quasi Experiment Design using pretest and posttest. The population of this study were students of class VIII Al Sudais Indonesia Foundation. The sample used was 2 classes selected by cluster random sampling technique, namely 1 class as an experimental class with the PMRI approach and 1 class as a control class with conventional learning. The data collection technique used is a test. Analysis of the data used is the normalize gain test, two-way ANOVA test, t-test, and Kruskal-Wallis test. The findings of this study are: a) the mathematical communication ability of students between those who take mathematics learning with the PMRI approach is higher than students who take conventional learning, b) the mathematical communication ability of male and female students between those who take mathematics learning with the PMRI approach is not. there are differences with students who follow conventional learning, and c) there is no interaction between the learning approach and gender differences in increasing mathematical communication skills