This study investigated the effects of Classwide Peer Tutoring (CWPT) on the reading skills of urban middle-school students using novels as the curriculum. Teacher-led instruction was compared with CWPT and CWPT plus a lottery contingency for appropriate on-task and tutoring behaviors. Three sixth-grade general education reading classes under the direction of one teacher participated. Data were collected on all students from weekly written tests of vocabulary and comprehension. Additional oral reading rate and academic engagement data were collected from three "low-achieving" target students. Overall, results demonstrated improved performance on weekly tests under CWPT conditions compared with teacher-led instruction. CWPT plus lottery resulted in further increases. Data also revealed differences in the types of academic responses made during teacher-led instruction and CWPT and increases in oral reading rates for two target students. These findings suggest that CWPT, particularly CWPT plus lottery, can improve the reading skills of urban middle school students. (Contains 4 figures and 1 table.)