School-effectiveness research has not yet been able to identify the factors of effective and noneffective schools, the real contribution of the significant factors, the true sizes of school effects, and the generalizability of school-effectiveness results. This paper presents findings of a meta analysis, the Dutch PSO programme, that was used to answer the above questions. The paper relates results to the size of reported school effects and to the effect size of the variable, educational leadership. Some persistent problems in the measurement of school effects are identified: specifically, measurement error, specification in relevant levels, and the choice of covariates. Studies conducted in the United States show a significant positive relationship between educational leadership and achievement, which raises the question why leadership is an important variable in the United States and not in other countries. Eight tables are included. Appendices contain statistical tables. (Contains 31 references.) (LMI)