When participants answer a test question and then receive feedback of the correct answer, studies have shown that the feedback is more effective when it is delayed by several seconds rather than provided immediately (e.g., Brackbill & Kappy, Journal of Comparative and Physiological Psychology, 55, 14-18, 1962; Schroth, Contemporary Educational Psychology, 17, 78-82, 1992). Despite several demonstrations of this delay-of-feedback benefit, a theoretical explanation for this finding has not yet been developed. The present study tested the hypothesis that brief delays of feedback are beneficial because they encourage anticipation of the upcoming feedback. In Experiment 1, participants answered obscure trivia questions, and before receiving the answer, they rated their curiosity to know the answer. The answer was then provided either immediately or after a 4-s delay. A later final test over the same questions revealed a significant delay-of-feedback benefit, but only for items that had been rated high in curiosity. Experiment 2 replicated this same effect and showed that the delay-of-feedback benefit only occurs when feedback is provided after a variable, unpredictable time duration (either 2, 4, or 8 s) rather than after a constant duration (always 4 s). These findings demonstrate that the delay-of-feedback effect appears to be greatest under conditions in which participants are curious to know the answer and when the answer is provided after an unpredictable time interval.