Many papers on the effect of single-sex education have been published. These studies on single-sex schools have two problems to be solved; 1) they focus on the effect on academic performance, and 2) they did not estimate with causal inference. In this paper, managing these two problems, we estimate the causal effect of the single-sex school on gender role attitudes with Japanese data. The data are nationally representative data for junior high and high school students, and this data contains information of respondents’ parents. To estimate the causal effect, we employ the IPW estimation method the propensity score with the information of parents. The treatment variable is attending single-sex schools, and the outcome variable is attitudes toward gender roles for child-bearing. The result shows that, contrary to the previous research, boys’ schools did not strengthen boys’ gender attitudes, while girls’ schools strengthen girls’ gender attitudes. This result is possibly led by girls’ oversocializetion in girls’ schools to the Japanese male-dominant society.