The present study investigated the effects of using different forms of material with 100 eleventh grade students enrolled in a 5-week CBI (computer based instruction) summer enrichment program in Memphis, Tennessee. The basic design consisted of two conditions of instructional support (text and questions vs. questions only), two testings (immediate vs. retention), five levels of similarity between lesson and posttest questions, and five feedback conditions: Knowledge of Correct Response (KCR), delayed KCR, Answer Until Correct (AUC), questions only (no feedback), and no questions. Results showed significant benefits of feedback over no-feedback, with AUC becoming more advantageous and delayed feedback less so as lesson-posttest question similarity decreased. Also, with decreased question similarity and the availability of supporting text, overall feedback effects tended to decrease. The results are discussed in terms of the information processing effects of the different feedback forms, a factor that CBI designers often fail to exploit in planning feedback conditions. Sample materials, data tables, and graphs are included. (41 references) (Author/BBM)