To evaluate different strategies for improving students' comprehension skills, particularly their inference ability, 42 fifth grade students identified as good or average readers were placed into one of three groups for an eight-week instructional period. Students in one group, the Content and Structure group, received systematic attention, story after story, to the importance of invoking background knowledge appropriate for story understanding and to eliciting an integrated understanding of the story in terms of salient components within a story grammar. The students in the second group, the Inference Awareness group, received treatment in which the process of drawing inferences from text and evaluating their plausibility was made explicit. Students in the third group, a placebo control group, participated in language activities unrelated to inferences. Results indicated that both experimental treatments were beneficial to students who, on one preexperimental measure or another, exhibited high aptitude for reading. The net result of the eight weeks of training was to show increasing gains as a direct function of how well students were already performing. Conversely, for students who performed relatively poorly on comprehension tasks prior to the experiment, the control group was either more or equally as beneficial. (FL)