This study was conducted to investigate the effects of motor imagery (MI) and mirror therapy (MT) on upper extremity function according to the level of cognition in stroke patients. Twenty-four participants who were diagnosed with stroke were divided into a mild cognitive group (n = 12) and a severe cognitive group (n = 12). Then, the two groups were again divided into MI group (n = 6) and MT group (n = 6). The participants were evaluated for function of upper extremity using the Box and Block Test (BBT), the Jebsen-Taylor Hand Function Test (JTHFT), and Manual Function Test (MFT). There were significant differences between the two groups of cognitive function of mild level in the post-test of JTHFT (p < 0.05). In the MI group, significant differences were found in the pre- and post-test scores for all variance (p < 0.05). In the MT group, significant differences were found in the pre- and post-test scores for JTHFT and MFT (p < 0.05). There were significant differences between the two groups of cognitive function of severe level in the post-test of all variances (p < 0.05). Furthermore, in the MT group, significant differences were found in the pre- and post-test scores for all variances (P < 0.05). The results of this study suggest that applying MI to the mild cognitive group is effective and that applying MT to the severe cognitive group is effective.