Background: Physical activity is an important contributor to reducing the risk for a variety of diseases. Understanding why people are physically active contributes to evidence-based planning of public health interventions because successful actions will target factors known to be related to physical activity (PA). Therefore the aim of this study is to identify the most meaningful correlates of PA in children and adolescents using a large, representative data set. Methods: Among n = 3539 (1801 boys) 6 to 17-year-old participants of the German representative Motorik-Modul baseline study (2003–2006) a total of 1154 different demographic, psychological, behavioral, biological, social and environmental factors were ranked according to their power of predicting PA using least absolute shrinkage and selection operator (LASSO) regressions. Results: A total of 18 (in girls) and 19 (in boys) important PA predictors from different, personal, social and environmental factors have been identified and ranked by LASSO. Peer modeling and physical self-concept were identified as the strongest correlates of PA in both boys and girls. Conclusions: Th ... mehre results confirm that PA interventions must target changes in different categories of PA correlates, but we suggest to focus particularly on the social environment and physical self-concept for interventions targeting children and adolescents in Germany nowadays. We also strongly recommend to repeatedly track correlates of PA, at least every 10 years, from representative samples in order to tailor contemporary PA interventions.