In a 3-year longitudinal study, the authors examined the effect of Direct Instruction (DI) on students' reading achievement, teacher perceptions, nature of the classroom, and special education referral rate. Urban and suburban students completed the Gates-MacGinitie Reading Tests in the fall of Grade 1 and spring of Grades 1, 2, and 3. Teachers completed interviews and questionnaires and were observed in their classrooms. We determined effects of reading instruction by covariate and correlation analyses. Results indicate that suburban low-skilled readers significantly benefited from the combination of DI and non-DI; urban students benefited from non-DI. Specific teacher characteristics significantly affected class reading achievement; DI did not decrease special education referral rates. (Contains 3 tables and 2 figures.)