In this paper I analyze the effect of physical education (PE) time on the body composition of students. Previous literature has established that spending more time with PE is largely ineffective in reducing children's body mass index (BMI). However, BMI can increase or can stay unchanged even when body composition becomes healthier, as physical activity can reduce fat mass and build muscle mass at the same time. Therefore, relying solely on BMI as an outcome can be misleading in assessing the true potential of PE time on students' health. In my analysis, along with BMI, I also use an outcome unavailable to previous large-scale studies, namely body fat percentage. Specifically, I analyze the effect of a Hungarian policy introducing daily PE classes for every student in the country in a phasing-in system starting from 2012. The policy created a large variation in PE time between subsequent age cohorts, which I can use for identification. Using data on the whole population of Hungarian 5th to 12th grade students I show that increased PE time was successful in reducing average body fat percentage among children, while at the same time it increased average BMI. I also show that despite the reduction in average body fat percentage, the prevalence of obesity among students stayed the same. Heterogeneity analysis reveals that increased PE time left obesity unchanged in specific subgroups of students (by age, income and gender) as well.