Purpose: To evaluate secular trends in recreational sedentary behavior among high school students in the United States between 2003 and 2015. Design: A series of cross-sectional assessments over a 12-year period. Setting: Data from the 2003 to 2015 Youth Risk Behavior Surveillance System was used. Participants: Samples of 10 978 to 14 894 adolescents, drawn every 2 years: 2003 to 2015. Measures: The evaluated recreational self-reported sedentary behaviors included TV hours and computer hours that are not schoolwork. Results: For the entire sample, and using polynomial orthogonal coefficients via regression modeling, there was an upward linear trend for total sedentary behavior hours (β = 0.03; p = .001), a downward linear trend in TV watching (β = −0.06; p < .001), and an upward linear trend in computer use (β = 0.08; p < .001) from 2003 to 2015. Similar linear trends ( p < .001) were observed across several subpopulations, including the groups by gender, race/ethnicity, and body mass index. However, various subpopulations differed in TV watching, with black or African American, and obese adolescents having the highest TV watching hours, respectively (eg, 3.82 h/d vs 3.13 h/d in 2015; blacks vs whites; p < .05). Various subpopulations also differed in computer use, with obese adolescents (4.26 h/d in 2015) having the highest computer use. Conclusion: There were significant changes from 2003 to 2015 in sedentary behavior patterns in the US adolescent population. Total recreational sedentary behavior increased in this period. Specifically, TV viewing decreased while computer use increased. Continued monitoring of sedentary behavior trends is needed to better understand the changing behaviors of American adolescents and how they relate to changes chronic disease risk.