OBJECTIVE: To evaluate the effect of computer-assisted cognitive training program (RehaCom®) on cognitive function of the patients with stroke. METHOD: Fifty seven subjects with stroke (34 males, 23 females) were enrolled and classified into two groups, experimental and control group. There was no significant difference between two groups in age, sex and lesion type distribution. Control group received conventional rehabilitation therapy including physical and occupational therapy. Experimental group received additional computer-assisted cognitive training using RehaCom software (Germany, 1996), 5 times per a week, 30 minutes per session, for 4 weeks. The RehaCom software consisted of reaction behavior, memory of words, topological memory programs. All patients were assessed their cognitive function using Computerized Neuropsychological Test (CNT), Lowenstein Occupational Therapy Cognitive Assessment (LOTCA) and Korean Version of Mini-Mental Status Examination (MMSE-K) before and after treatment. Functional independence measurement (FIM) was also applied for evaluation of functional status. RESULTS: There was no difference between two groups in the LOTCA, CNT and FIM scores at baseline. Four weeks later, scores of the MMSE and FIM were significantly improved in the experimental group compared to the control group (p<0.05). Especially, the improvement was significant in moderate cognitive impairment group (MMSE = 11~21) (p<0.05). In learned patients of experimental group, the score of the MMSE and LOTCA were significantly more improved than control group (p<0.05). CONCLUSION: Computer-assisted cognitive training would be useful as a additional tool of cognitive rehabilitation in patients with stroke. Especially, the effect of computer-assisted cognitive training program was far better in patients with moderate cognitive impairment and in patients who show learning in cognitive training program.