Sixty undergraduate students read and summarized eight short passages in a study of how semantic components were involved in text comprehension. After the first summary was completed (the naive or untutored summary), the students were assigned to one of three groups. The experimental group was introduced to the semantic components, including agent, event, patient, instrument, result, and descriptive. One control group was tutored in a 'traditional' method, stressing 'topics' and 'main points.' The second control group was untutored throughout the task. When the naive summaries were evaluated, the better summaries contained significantly higher proportions of the semantic component information. Analyses of variance showed that the experimental group summaries were significantly better than the summaries by the traditional and the untutored groups. It seemed clear that the 'natural' categories of semantic components identified certain types of judgments that were more accessible to processing needs than the less natural concepts of 'topic,''main points,''grammatical subject,' and 'direct object.' Thus, instruction in using semantic categories helped readers sift more efficiently through the potential information in a passage, allowing them to integrate the important information more completely. (RL)