This systematic review presents the first synthesis of the scientific literature on estimating school and teacher/class effects on student academic performance using random-effects (RE) models with three or more levels. The review delves into the theoretical framework underpinning the estimation of educational effects, the associated statistical methods, and the empirical findings from multilevel analyses to date. It includes 94 analyses in 32 papers from diverse geographical regions, focusing on both primary and secondary education levels. The findings reveal a broad spectrum of Variance Partition Coefficients (VPCs) from RE models, highlighting the variability in the impact of school and teacher/class levels on student outcomes. The review emphasizes the necessity for comprehensive and diverse datasets, particularly the inclusion of detailed teacher-level data in African studies and the utilization of robust official educational databases in Chinese research. It also acknowledges the contributions of Latin American countries in advancing multilevel modeling and the holistic approach to educational analysis demonstrated in Cyprus. The conclusions underscore the importance of contextualizing educational policies and practices based on the multifaceted influences on student performance. The review calls for future research to expand data collection efforts, adopt standardized reporting practices, and explore advanced statistical techniques to address methodological challenges and approximate causal relationships more accurately.