A study was conducted to test the findings of two earlier studies (Peeck l974 and Pressley l983) on the effects of occasional mismatches between verbal and pictorial content in children's retention of illustrated prose. While the Peeck study indicated a considerable impact of mismatched pictures, the Pressley study indicated that with some procedural changes in the design of mismatched picture-experimentation, no reduction of children's retention of prose will occur. In the present study, subjects (53 fifth grade children) were assigned to one of four conditions defined by the presence or absence of illustrations. The text was a 19-page story with 37 illustrations. Subjects were told to select the alternative corresponding to the text each time they came across a question dealing with a mismatched illustration. Retention was measured either by a multiple-choice test or by cued recall. Results suggest substantial effects of mismatched pictures, especially when retention is assessed after a delay. Effects are due to the forgetting of initially detected inconsistencies, confusion of the origin (verbal or pictorial) of retained information, and an increased tendency to respond with information provided by pictures rather than by text. (EL)