Teachers' expectations of student capabilities can meaningfully affect future achievement. However, the consistency of teachers' expectations over time and their differential impact by student gender and ethnicity have not been thoroughly investigated in the context of students' reading achievement. The current study investigates the longitudinal influence of teacher expectancy effects on student achievement through a novel measurement approach that considers classroom context while drawing on cognitive assessments and teacher ratings collected through the large, nationally representative Early Childhood Longitudinal Study-Kindergarten data set (N = 8,503 students). Results of the autoregressive cross-lagged models with five time points between kindergarten and eighth grade indicate teachers' expectations of students' reading achievement are not significantly associated from one time point to the next, though their effects increase over time. Significant differences in longitudinal expectancy effects based on student gender and ethnicity were not found. Implications for teaching practice based on these results are discussed.