This conceptual paper addresses fluency instruction and assessment for students with or at risk for reading disabilities. Although a multidimensional construct, fluency tests more often have been limited to measuring students' reading rate and accuracy. The reasons for this are explained as well as how fluency tests have influenced classroom instruction in ways not supported by research. Specifically, an argument is presented against using fluency scores to restrict the texts that students are assigned to read. Then, the basis for having students repeatedly read a single text is described along with literature suggesting this common practice may not lead to expert reading performance. Finally, the paper presents recommendations for forms of fluency instruction that are supported by research as well as a promising practice aligned to those recommendations, called Varied Practice Reading (VPR). The basis of VPR, its components, and the early results of implementation studies in Grades 4 and 7 also are described.