As compared with providing extrinsic feedback on each of a set of practice trials, reducing the feedback frequency in various ways facilitates long-term retention. One explanation is that frequent feedback operates proactively on the subsequent trial, inducing excessive variability that degrades learning. We tested this view by giving or not giving feedback "reminders," where the postresponse feedback was given again just before the next attempt at a task. The reminder manipulation was examined in both blocked and randomized practice sequences during the learning of three limb-patterning tasks. Reminder feedback increased response variability during practice in both random and blocked practice. As measured in retention tests, though, feedback reminders degraded learning in random practice, but not in blocked practice. This implies that the frequently found learning advantages of random, as compared with blocked, practice might be due to feedback's facilitation of retrieval operations in blocked practice (as well as in random practice with feedback reminders); such overfacilitation of retrieval operations has been shown to degrade learning. Additional analyses revealed that random, as compared with blocked, practice enhanced the learning of the fundamental pattern of action (generalized motor program) but had little effect on the ability to scale the pattern in amplitude or time (parameterization).