This study compared the effects of two types of delayed feedback (correct response or correct response + rationale) provided to students by a computer-based testing system following an exam. The preclinical medical curriculum at the University of Kansas Medical Center uses a two-exam system for summative assessments in which students test, revisit material, and then re-test (same content, different questions), with the higher score used to determine the students' grades. Using a quasi-experimental design and data collected during the normal course of instruction, test and re-test scores from midterm multiple choice examinations were compared between academic year (AY) 2015-2016, which provided delayed feedback with the correct answer only, and AY 2016-2017, where delayed feedback consisted of the correct answer plus a rationale. The average increase in score on the re-test was 2.29 ± 6.83% (n = 192) with correct answer only and 3.92 ± 7.12% (n = 197) with rationales (p < 0.05). The effect of the rationales was not different in students of differing academic abilities based on entering composite MCAT scores or Year 1 GPA. Thus, delayed feedback with exam question rationales resulted in a greater increase in exam score between the test and re-test than feedback with correct response only. This finding suggests that delayed elaborative feedback on a summative exam produced a small, but significant, improvement in learning, in medical students.