Two experiments explored the benefits to retention and transfer conferred by mental practice. During familiarization, participants typed 4-digit numbers and took an immediate typing test on both old and new numbers. Participants then typed old 4-digit numbers, either physically or mentally, with either a different response configuration or the opposite hand from that used during familiarization. On a delayed test, participants physically typed both old and new numbers with the same response configuration and hand used during familiarization. Mental practice led to less retroactive interference and more transfer than did physical practice, supporting the hypothesis that mental practice strengthens an abstract representation that does not involve specific effectors. (Contains 2 figures, 1 footnote and 6 tables.)