This study aimed to overcome gaps in the literature by investigating effects of EMG feedback provided specifically from spastic muscles during antagonist phases of active movement, while controlling for the effects of practicing the movement and while testing for transfer of training without confounding several transfer phenomena. In a single training session, two groups of stroke patients practiced a pursuit tracking task by following a moving target with elbow flexion and extension, simultaneously attempting to reduce the activity of elbow flexors. Both groups tracked the target more accurately following training. Transfer tests failed to demonstrate effects of feedback on accuracy of tracking or on electromyographic activity during performance of the practiced task without feedback. Moreover, the group that was trained with electromyographic feedback exhibited negative transfer on variants of the practiced task: tracking faster or less predictable targets.