BACKGROUND AND PURPOSE: After a stroke, people often attempt to consciously control their motor actions, which, paradoxically, disrupts optimal performance. A learning strategy that minimizes the accrual of explicit knowledge may circumvent attempts to consciously control motor actions, thereby resulting in better performance. The purpose of this study was to examine the implicit learning of a dynamic balancing task after stroke by use of 1 of 2 motor learning strategies: learning without errors and discovery learning. PARTICIPANTS AND METHODS: Ten adults with stroke and 12 older adults practiced a dynamic balancing task on a stabilometer under single-task (balance only) and concurrent-task conditions. Root-mean-square error (in degrees) from horizontal was used to measure balance performance. RESULTS: The balance performance of the discovery (explicit) learners after stroke was impaired by the imposition of a concurrent cognitive task load. In contrast, the performance of the errorless (implicit) learners (stroke and control groups) and the discovery learning control group was not impaired. DISCUSSION AND CONCLUSION: The provision of explicit information during rehabilitation may be detrimental to the learning/relearning and execution of motor skills in some people with stroke. The application of implicit motor learning techniques in the rehabilitation setting may be beneficial.