To test the clinical counterpart of the learned nonuse theory, 25 chronic hemiplegic stroke and head-injured patients with minimal to moderate upper extremity extensor muscle function were required to keep their uninvolved upper extremities within a hand-enclosed sling during waking hours over a 2-week interval. During this forced use period and for 1 year thereafter, changes in force or time-based measures among 21 functional tasks were compared to values at the sixth baseline session, a preintervention time when relearning had plateaued. Significant (P less than 0.05, Friedman's repeated measures followed by Tukey multiple comparison tests) changes were seen in 19 of the 21 tasks with most persisting at the 1-year follow-up. There were no apparent differences between right- and left-sided involvement or between stroke versus head injury clients (Mann-Whitney procedure). Ratings for quality of movement scored from videotapes presented in random order showed no change over time. These data suggest that learned nonuse does occur in select neurological patients and that this behavior can be reversed through application of a forced use paradigm.