Developmental differences in summaries of expository texts were examined for fourth and sixth grade students in two experiments. Experiment 1 used texts that had a description content structure whereas Experiment 2 texts had a thesis/evidence content structure. Manipulation of the presence or absence of the topic sentence allowed the examination of the developmental relationship between selection and invention strategies. Processing demands were manipulated by readability level: each group received texts at and 2 years below their grade level, as measured by the Fry Readability formula. Each subject constructed summaries of three different lengths. In both experiments, differences between fourth and sixth graders' summaries were most apparent at the shorter summary lengths. Grade differences were larger when the topic sentences were not in the texts. The obtained pattern of effects is consistent with the notion that more sophisticated summarization strategies place greater information processing demands on children. (Author/PN)