Reading instruction and its effects were examined for 105 students in elementary classrooms for the learning-disabled. Extensive detailed observations of students, teachers, and instructional material were used to explore the plausibility of a causal model of the effects of reading behaviors and instruction on students' reading performance. The results indicate that 72 percent of the variance in posttest reading scores can be explained by a model that includes a pretest, three student reading behaviors, and instructional overlap; and that 59 percent of the variance in student time spent in reading can be explained by a model that includes pretest, teacher instructional behaviors, teacher affective behaviors, and instructional pacing. Using these results as a basis, mechanisms for increasing salient student reading activities in classrooms that serve students with reading problems are discussed.