It was hypothesized that the cognitive structures which underlie performance on class-inclusion tasks also underlie reading comprehension of the stated or inferred main idea of a paragraph, and that performance on three-dimensional class-inclusion tasks developmentally precedes performance on two-dimensional class-inclusion tasks. Ten kindergarten through fourth grade students were administered two comprehension tests and eight class-inclusion tasks in the pilot study, which is appended. As a result, the Main Idea Test (MIT) and seven single and double class-inclusion tasks were administered to 267 kindergarten through second grade students. The MIT involves three paragraph types: (1) those with the main idea stated in the first sentence; (2) those with the main idea in the last sentence; and (3) those with the main idea to be inferred by the reader. Each paragraph is followed by a multiple choice question on the main idea. Kindergarten students scored at a chance level on the MIT. Level of class-inclusion was significant for first and second graders. Second graders scored significantly lower on inferred than on stated main ideas. Implications for further research are discussed, and MIT items are appended. (Author/JAC)