The aim of this study was to compare the learning process of a postural control task between post-stroke patients and healthy subjects. The sample was composed of 20 post-stroke individuals (Experimental Group) and 20 aged matched healthy individuals (Control Group). Participants practiced a postural control task in a virtual environment with increasing of complexity. The study design involved four phases: pre-test (five trials), acquisition phase (four blocks of thirty minutes), post-test (five trials), and retention test (five trials after a week without practice). The statistical analysis was run by a 2 x 3 ANOVA (groups x learning tests). Results: There was no difference in motor learning between Experimental Group and Control Group (F= 41.22; p=0.88). In addition, it was founded that the Control Group could learn the task in a higher-level complexity than Experimental Group (F = 4.77; p = 0.01), and both groups increased the error during the trials of practice (F = 0.53; p = 0.00) because of task complexity.  Conclusion: Therefore has been found that post-stroke individuals have the ability to learn a postural control task similar to healthy subjects, and the task complexity seems to be a key-factor in order to differentiate stroke from healthy subject's motor learning process.