High and low level questions as determined by a panel of evaluators were combined with corrective feedback and attribute isolation feedback to form four versions of a computer-based science lesson. The sample consisted of 154 high school chemistry students in a suburban high school. The primary hypothesis was that students who received high level questions and attribute isolation feedback would perform significantly higher on posttests than students who received the other combinations of questioning and feedback. The experiment used the three-factor, repeated measures design known as the two between-one within-subjects design defined by between-subjects variables A (levels of questioning) and B (feedback conditions), which cross, and by the within-subject factor C (scores on Posttest I and Posttest II) treated as a repeated measurements factor. Results indicate the primary hypothesis was not sustained, but the question-level by posttest interaction and the question-level and posttest main effects were statistically significant at the .05, .001, and .01 levels respectively. The question-level main effect and the question-level by posttest interaction provided verification of the efficacy of levels-of-questions as an instructional factor in courseware design. High level questions were more likely to cause deep cognitive processing and to enhance retention than low level questions, and should be included appropriately in the design of computer-based instructional systems. A four-page list of references is provided. (Author/JB)