Three experiments were conducted, all employing undergraduates in college courses taught according to personalized system of instruction (PSI) principles. Experiment I examined retention as a function of the feedback delay interval in an introductory anthropology course using short-answer essay tests. Experiment II varied the feedback delay interval, the informational quality of the feedback, and test item type, and Experiment III examined delay and item type in a psychology course on experimental design. Results showed that neither delayed nor immediate feedback produced superior, long-term knowledge retention, regardless of feedback quality. Further, subjects in the immediate feedback conditions did not repeat initial errors more frequently than did delay subjects. The frequent, repeatable-quizzing aspect of PSI may make feedback a less potent variable than in other courses, and results do not support the use of immediate feedback when cost considerations and convenience or administration are important. Fourteen references and 16 data tables are included. (LMM)