Recent studies (Chiviacowsky & Wulf, 2002, 2005) have shown that learners prefer to receive feedback after they believe they had a "good" rather than "poor" trial. The present study followed up on this finding and examined whether learning would benefit if individuals received feedback after good relative to poor trials. Participants practiced a task that required them to throw beanbags at a target with their nondominant arm. Vision was prevented during and after the throws. All participants received knowledge of results (KR) on three trials in each 6-trial block. While one group (KR good) received KR for the three most effective trials in each block, another (KR poor) received feedback for the three least effective trials in each block. There were no group differences in practice. However the KR good group showed learning advantages on a delayed retention test (without KR). These results demonstrated that learning is facilitated if feedback is provided after good rather than poor trials. The findings are interpreted as evidence for a motivational function of feedback.