This study assessed the predictive power of variables at school-level and student-level on the risk of school dropout within the framework of the Ecological Theory. A predictive correlational research model was used, and the participants were selected by employing a two-stage sampling procedure. Participants were 1.851 students from 30 different high schools in the central districts of Ankara studying during the 2016/17 school year. According to the findings of two-level hierarchical linear regression analysis; the average of the highest school at the risk of school dropout is almost twice that of the lowest school. The effect of the environmental risk perception on school dropout in schools with high dropout is higher than schools where the risk of dropout is low. Academic achievement, environmental security, and non-antisocial behaviors reduce the likelihood of early school leaving at the schools with a higher risk of dropout than lower-ones. The probability of dropout of upper grades in schools comprising student clubs and students with high academic achievement in schools with High School Entrance Exam success is less. It was also concluded that the effects of gender, class repetition, having a sibling who left school early, teacher participation, punishment, absenteeism and motivation may have been mediated by other variables in the regression equation.