OBJECTIVE: Previous studies in rehabilitation in poststroke hemiplegia have suggested that brief periods of bilateral training improve subsequent unilateral performance of the paretic upper limb. The bilateral training protocol involved task-specific practice using both upper limbs simultaneously and homologously (such that they mirrored each other). This study aimed to determine the generalizability of the bilateral training phenomenon by replicating the work with different tasks, measures, and a different experimenter. METHODS: Five single-case experiments were performed using a within-participant multiple-baseline design. Participants with post-stroke hemiplegia underwent 25 to 40 daily training sessions involving repetitive practice of 3 copying tasks. Pen movements were recorded on a digitizing pad. During the baseline phase, participants performed tasks unilaterally with their hemiplegic limb. Bilateral therapy was then applied in a staggered fashion to each task in turn. Quality of movement was measured in terms of temporal performance (jerkiness, speed, task duration), spatial performance (accuracy of copy), and arm posture (pen tilt). RESULTS: Movement quality of the hemiplegic limb was not improved during or subsequent to regular bilateral training. CONCLUSIONS: The training phenomenon observed in previous work was not replicated when participants with hemiplegia practiced copying tasks bilaterally.