To determine whether poor readers can predict as well as good readers when they read easy material and to determine the effect of material difficulty on predictive ability, a study involving 140 tenth and eleventh graders was conducted. The subjects were randomly selected, with 70 good readers chosen from students who scored at stanines 7, 8, or 9 on the "Stanford Diagnostic" Reading Test and 70 poor readers drawn from those who scored at stanines 2, 3, or 4. Each subject read two short stories which had been selected according to length, readability, and writing style. Each story was divided into five sections; multiple choice questions accompanied each section. After every two sections a page was inserted for the subject to write what would occur next. Results showed that good readers have more accurate predictions than poor readers and that readers have more accurate predictions with concrete-style material than with abstract-style material. Further research should investigate the process by which a reader can predict the author's syntax or content. Research should also investigate whether prediction is a necessary process for comprehension. If it is, ways of teaching it should be explored. (FL)