Learners ( n = 48) practiced three multisegment movements with distinct target movement times. Four practice groups were compared: blocked, random, and two groups who had a win-shift/lose-stay schedule (WSLS1 and WSLS2). For these latter groups switching between practice tasks was performance-contingent: within 5% of target time for 1 or 2 consecutive trials, respectively. During acquisition, blocked performance was more accurate than for both random and WSLS2 groups. The WSLS1 group performed between blocked and random groups, but did not differ from either. In a next-day retention test, the random group scored better than the blocked group. The WSLS1 group performed similarly to the random practice while the WSLS2 group's scores were similar to those of the blocked group. Results encourage further study of similar practice schedules.