Impaired locomotion in individuals with post-stroke hemiplegia is due, in part, to abnormal muscle activation patterns with respect to amplitude and phasing. It can be theorized that interventions that improve muscle activation patterns may decrease disability in persons with hemiplegia following stroke. The purpose of this study was to determine the change in muscle activation patterns of three lower extremity muscles during forward pedaling after 15 minutes of backward pedaling training on a modified ergometer in persons with post-stroke hemiplegia. Previous studies have shown that during backward pedaling, there is a large reduction in inappropriately prolonged vastus medialis (VM) and soleus (SL) activity that is present during forward pedaling. Therefore, we hypothesized that, after 15 minutes of backward pedaling, inappropriate activation of the VM and SL during foward pedaling would be reduced. In contrast, since biarticular muscles, particularly the rectus femoris (RF), do not show altered phasing during backward pedaling, we hypothesized that the RF would not change its inappropriate phasing during forward pedaling after backward pedaling training. Eleven participants with a single unilateral stroke and lower extremity paresis greater than six months duration were tested. EMG was recorded for the paretic and nonparetic lower extremity muscles as a 1-minute baseline measurement of muscle activation patterns during forward pedaling. Participants performed both a 15-minute bout of backward pedaling and a 15-minute bout of forward pedaling followed by a post-training 5-minute bout of forward pedaling. Contrary to expectations, inappropriate VM activation in the paretic lower extremity was increased rather than reduced during forward after backward pedaling training. Consistent with our hypothesis, inappropriate RF phasing did not change after backward pedaling training. We conclude that, even though muscl activity patterns are changed with changes in task mechanics, those chnges are not transferable to the original task. In addition, there may be some evidence to support inappropriate phasing enhancement training to reduce post-training error.