Purpose: The purpose of this study was to examine changes in school-based physical education (PE) attendance over time among nationally representative samples of U.S. high school students and how changes in PE attendance have varied across demographic subgroups. Method: Student demographic information and PE attendance data were obtained from 13 biennial cycles (1991-2015) of the national Youth Risk Behavior Survey (YRBS). Physical education variables derived from YRBS data included "PE attendance," "daily PE," "average PE days/week," and "PE frequency." Logistic regression models examined trends in "PE attendance," "daily PE," and "PE frequency" for the overall sample and demographic subgroups. Linear regression models examined trends in "average PE days/week" in the overall sample only. Results: Overall, there was no significant change in the percentage of students reporting "PE attendance" during 1991 to 2015. However, "daily PE" and "average PE days/week" declined significantly from 1991 to 1995 (41.6% to 25.4% and 4.64 days to 3.64 days, respectively) and then remained stable through 2015 (29.8% and 4.11 days, respectively). The percentage of students reporting a "PE frequency" of 3 days per week increased significantly from 1991 to 1995 (1.5% to 19.0%) before stabilizing through 2015 (9.1%). Trends across demographic subgroups revealed notable differences in PE attendance. Conclusions: Study findings showed that U.S. schools have not substantially reduced PE amounts in recent years. Still, the prevalence of PE attendance among U.S. high school students is well below recommendations. For PE to contribute to increased adolescent compliance with national physical activity guidelines, significant policy actions are needed to improve PE access for all students.