This study compared a multiple exemplar approach involving the training of key words and sentence structures to a typical repeated readings procedure for their effects on students' generalized oral reading fluency. The two training approaches were also compared in terms of their relative learning rates (i.e., fluency gain per minute of instruction time). Participants were 111 first- and second-grade students from three public schools in a large urban school district. This study utilized an across students pre-test/post-test experimental design and one-way analyses of variance on the gain scores from pre-test to post-test to assess the effectiveness of multiple exemplar instruction, listening passage preview/repeated readings, and a time and attention control condition. Results showed significantly higher gain scores for students in both the multiple exemplar and listening passage preview/repeated readings groups versus controls on the trained and generalization passages. Multiple exemplar instruction also resulted in significantly larger learning rates than listening passage preview/repeated readings. Implications, limitations, and directions for future research are discussed.