Two important components of Constraint-Induced Movement therapy are thought to be intense training of the more-impaired arm and physical restraint of the less-impaired arm. This preliminary study examined the effects of type of training (task-practice, shaping) and restraint (sling, half-glove, no restraint) on treatment outcomes. Seventeen individuals at least 1-year post-stroke with mild/moderate upper extremity motor deficit were consecutively assigned to Sling and Task-practice, Sling and Shaping, Half-glove and Shaping, and Shaping Only groups. Task-practice involved repetitive more-impaired arm training on functional tasks for 6 hr/day for 10 consecutive weekdays. Shaping differed from task-practice in that task demands were progressively increased and immediate performance feedback was provided frequently and systematically. "Sling" groups placed the less-impaired arm in a resting hand-splint/sling assembly for most waking hours over the 2-week intervention, while the "Half-glove" group wore a modified gardening glove as a reminder not to use the more affected arm in the life situation. There were no between-group differences in outcome at post-treatment, although two-years afterwards Sling & Task-practice and Half-glove & Shaping participants showed larger and smaller retention of gains, respectively, than those in the Sling & Shaping group. Thus, long-term outcomes may have been affected by type of more-impaired arm motor training and less-impaired arm restraint. These variables, however, were confounded with between-group differences in training intensity, limiting confidence in this conclusion. © 2006 - IOS Press and the authors. All rights reserved.