A series of studies with adults resulted in a standardized test to assess children's comprehension of main ideas in simple narratives. The test, utilizing stories from the WISC-R picture arrangement task, was then employed in a developmental investigation with second, fifth, and eighth grade students. Each student completed the WISC-R arrangement task in its usual form in one session and ranked main idea alternatives associated with the picture stories in a second. In addition, a standardized measure of reading comprehension was available for most students. Results showed a clear improvement across the grades in children's ability to select the best topic statement for the picture stories. The high consensus alternative, which abstracted the main action sequence, was perceived as best. The single picture alternative, which stated a specific action tied to one picture in the sequence, was perceived as second best. The setting alternative, which described some feature of the story's context or background, was perceived as the third best alternative. The wrong alternative, which stated an action inappropriate to the story, was perceived as the worst alternative. Developmentally, the younger children seemed to have more difficulty appreciating or detecting the "completeness" of action stated in the alternative and using such a criterion to discriminate the quality of the topic statement. (Author/HOD)