The ability to correctly answer reading comprehension test items, without having read the accompanying reading passage, was compared for third grade learning disabled students and their peers from a regular classroom. In the first experiment, fourteen multiple choice items were selected from the Stanford Achievement Test. No reading passages were provided, but the items were clustered according to their association with a particular passage. The learning disabled students answered correctly only slightly more than predicted by chance. Scores of the non-learning disabled group were significantly higher. In order to address the suggestion that the poorer performance of learning disabled students was caused by their lesser ability to read the test questions, a second experiment was conducted. A different sample of students included learning disabled and average third grade students. The examiner read each of the 14 test items aloud, and all students were given sufficient time to answer all items. The average students again had higher scores than the learning disabled group. The results suggested that learning disabled students may be less likely than their non-learning disabled peers to apply test-taking strategies to reading comprehension questions. (GDC)