A study examined the effects of prediction training on the reading comprehension and written composition performance of fourth-grade students on the following reading and writing tasks: story recall, story generation, and number of relevant predictions. Subjects, 40 students attending two private urban elementary schools, were assigned to one of two treatment conditions (prediction training or rereading) and each group participated in three training sessions. Results showed that students who received prediction training outperformed the reread group on generating predictions about stories they were reading and in story generation. No significant differences were found between the groups with respect to free and cued recall performance following story reading. (One table of data is included and 25 references are attached.) (SR)