Imagery practice of motor tasks has been recommended for a wide range of activities as from flight training to basketball. A key question, both from a practical and a theoretical standpoint, is when during the learning process does imagery practice confer the most benefit? However, the literature does not provide clear guidance, in part because of methodological limitations. A 3 x 2 (physical practice x pretest-posttest) split-plot design was employed to investigate the effects of imagery practice on the acquisition of a discrete target at three different stages of learning. Analysis indicated that imagery practice was most beneficial in the early stages of learning and showed an inverse relationship between experience and efficacy of imagery practice. Results are discussed in terms of current theories of imagery practice and suggestions are made regarding when such practice might be best applied during skill development.