Objectives. Sensory and motor alterations resulting from stroke often impair the performance and learning of motor skills. The present study is aimed at investigating whether and how poststroke individuals and age- and sex-matched healthy controls benefit from a contextual interference effect on the practice of a maze task (i.e., constant vs. random practice) performed on the computer. Methods. Participants included 21 poststroke individuals and 21 healthy controls, matched by sex and age (30 to 80 years). Both groups were divided according to the type of the practice (constant or random) presented in the acquisition phase of the learning protocol. For comparison between the groups, types of practice, and blocks of attempts, the analysis of variance with Tukey’s post hoc test (p<0.05) was used. Results. Poststroke individuals presented longer movement times as compared with the control group. In addition, only poststroke individuals who performed the task with random practice showed improved performance at the transfer phase. Moreover, randomized practice enabled poststroke individuals to perform the transfer task similarly to individuals without any neurological impairment. Conclusion. The present findings indicated a significant effect of contextual interference of practice in poststroke individuals, suggesting that applying randomized training must be considered when designing rehabilitation protocols for this population.