Despite many previous studies on reading fluency (measured by a maze task) as a screening measure, our understanding is limited about the utility of silent reading fluency in predicting later reading comprehension and contextual influences (e.g., schools and districts) on reading comprehension achievement. In the present study we examined: (1) How much variance in reading comprehension scores exist between students, classes, schools, and districts for children in grades 3-10; and (2) whether silent reading fluency measured by a maze task adds a unique contribution to the prediction of spring reading comprehension after accounting for fall spelling and reading comprehension. Results showed that a substantial amount of variance in reading comprehension is attributable to differences among classrooms (21-46%), particularly in grades 6-10. In addition, approximately 3-5% of variance in reading comprehension was attributable to differences among schools and districts. Silent reading fluency also explained a unique amount of variance in spring reading comprehension after accounting for students' performance in reading comprehension and spelling in the fall. Unique variance (pseudo-"R" [superscript 2]) varied from 2 to 10% at the student, class, school, and district levels. These results suggest that a maze task has potential utility as a screening measure of reading comprehension for students in grades 3-10. Furthermore, differences among classrooms, schools, and districts matter for students' reading comprehension achievement.