Reading instruction often does not focus on appealing to student interests, offering choice, or responding to the needs of advanced readers. In this experimental study, we examined the effects on achievement of an instructional approach involving choice, differentiated instruction, and extensive, supported, independent reading, with corresponding elimination of regular reading instruction. The study, which incorporated multi-site cluster-randomized design, was conducted in four middle schools with 2,150 students and 47 teachers. Pretest and posttest data were collected on reading fluency and comprehension, with Hierarchical Linear Modeling (HLM) procedures used to investigate the effects of the intervention. Results indicated similar results overall for treatment and control group students, with treatment outperforming control on reading fluency at two of the schools. The findings demonstrate that the intervention resulted in similar or higher scores for fluency and similar scores for comprehension, despite the diminished whole-group and small-group instruction provided in the intervention as compared with regular reading classes.