This study uses data from a comprehensive database of U.S. early childhood education program evaluations published between 1960 and 2007 to evaluate the relationship between class size, child-teacher ratio, and program effect sizes for cognitive, achievement, and socioemotional outcomes. Both class size and child-teacher ratio showed nonlinear relationships with cognitive and achievement effect sizes. For child-teacher ratios 7.5:1 and lower, the reduction of this ratio by one child per teacher predicted an effect size of 0.22 standard deviations greater. For class sizes 15 and smaller, one child fewer predicted an effect size of 0.10 standard deviations larger. No discernible relationship was found for larger class sizes and child-teacher ratios. Results were less clear for socioemotional outcomes due to a small sample. (author abstract)