Results of Leinhardt, Zigmond, and Cooley (1981) have been interpreted as support for increased silent reading in classroom reading instruction. Leinhardt et al. examined a causal model of classroom processes influencing reading achievement and showed that time spent in silent, rather than oral, reading was positively related to gains in reading achievement. The present study reanalyzed the Leinhardt et al. data using linear structural equation modeling. We show that students’ entry-level reading abilities had a significant direct effect on time spent in silent reading but no such effect on time spent on oral or “indirect” reading. When entry-level abilities were more adequately controlled, silent reading no longer showed a significant effect on posttest reading performance. Under alternative models of the data, there is even the suggestion that time spent in oral reading had more effect on final reading achievement.