A path model of teacher expectancy effects was evaluated in a sample of 376 first- through fifth-grade urban elementary school children. The roles of two moderators (classroom perceived differential treatment environment and developmental differences) and one mediator (children's self-expectations) of teacher expectancy effects on children's year-end achievement were examined. Significant differences in effects and effect sizes are presented. Both classroom environment (high versus low in differential treatment, as seen through children's eyes) and developmental differences moderated the strength of teacher expectancy effects. Generally, stronger effects were found in classrooms in which expectancy-related cues were more salient to children, but developmental differences moderated which effect was most pronounced. A significant age-related decline in direct effects on ending achievement was interpreted as evidence that teacher expectations may tend to magnify achievement differences in the early grades, but serve to sustain them in later grades. Support for indirect effects (teacher expectations [RIGHTWARDS ARROW] children's self-expectations [RIGHTWARDS ARROW] ending achievement) was limited to upper elementary grade classrooms perceived as high in differential treatment. In contrast to prior research that emphasized small effect sizes, the present analyses document several instances of moderate effects, primarily in classrooms in which expectancy-related messages were most salient to children. These results underscore the importance of explicit attention to the inclusion of moderators, mediators, and multiple outcomes in efforts to understand teacher expectancy effects.