When students are tested, delaying feedback of the correct answer by a few seconds is more beneficial than providing it immediately (e.g., Brackbill & Kappy, 1962). The current study refers to this effect as the delay-of-feedback benefit. Little is known about the theoretical nature of the delay-of-feedback benefit. The current study investigated the hypothesis that a delay of feedback is beneficial because it allows a learner to anticipate--i.e., look forward to, or devote attentional resources to--the feedback when it arrives. Participants were asked a list of obscure trivia questions. Immediately after answering each question they were asked to indicate how curious they were to know the answer (on a scale of 1-6), which was then presented to them immediately or after a delay of a few seconds. A final test given several minutes later revealed better memory for items that had previously received delayed feedback compared to immediate feedback, but only under conditions in which participants expressed high curiosity to know the answer. No delay-of-feedback benefit ever occurred for items that received low curiosity ratings. This interaction between feedback timing and curiosity was obtained in Experiment 1 in which immediate vs. delayed feedback was manipulated within-subjects. Experiment 2 manipulated feedback timing between-subjects and obtained this same interaction only when the duration of the delay was varied (i.e., either 2, 4, or 8 seconds) rather than constant (i.e., always 4 seconds). These results demonstrate that participants are most likely to benefit from delayed feedback when they are curious about the answer, and when they are uncertain about exactly when the answer will appear.