A study of the spacing effect was conducted within a short-term motor memory framework. In Experiment 1 subjects received one or six repetitions of a discrete linear movement at intervals of 5 or 60 sec, and with a constant retention interval of 30 sec. The performance over 10 trials was analyzed in terms of algebraic, absolute, and variable error. The facilitation associated with longer intervals found in verbal retention situations was obtained in the present study for algebraic error only. The results were discussed in terms of the several hypotheses offered by Hintzman (1974), and a mechanism for the operation of differential attention was suggested. Inconsistencies with previous data were discussed. Experiments 2 and 3 addressed some unresolved issues from Experiment 1 and additional support for the differential-attention hypothesis was obtained.