School autonomy has been and continues to be one of the most important education reform strategies around the world despite ambiguity about its theoretical and empirical effects on students learning. We use international data from PISA to test three country-level factors that might account for inconsistent results in prior literature: (1) the selective implementation of school autonomy based on school performance; (2) differential influence on high-risk subgroups; and (3) the presence of accountability policies to prevent opportunism by autonomous schools. We find that the relationship between autonomy and student test performance varies both across countries and within countries across subgroups in both magnitude and direction. Similar results are observed if decentralization is coupled with accountability policies. All of three tested factors influence country-level associations between school decentralization and student learning, which suggests that autonomy is effective only when contextual factors and other policies are aligned.