A study investigated the hypothesis that mental imagery facilitates access to prior knowledge and therefore enhances the ability to infer and make text-relevant predictions. The subjects, 29 first grade and 29 third grade students, were randomly assigned to either an experimental group where they were instructed to make pictures in their head to help them remember the story as they read, or a control group where they were told simply to think about what they were reading to help them remember. A short story, which was written at first grade and third grade readability levels, was divided into five sections at points in the story where predictions about forthcoming events could be made and used as stimuli. Immediately following the silent reading of each of the five sections of the story, the subjects responded to a prediction question. The recorded responses were scored for reference to explicit facts, prediction statements, and accuracy of prediction. The results supported the hypothesis that induced mental imagery enhances the ability to infer and make text predictions for third grade students, but not for the younger subjects. (HTH)