OBJECTIVE: To assess the properties of the Mandarin version of the Stoke Impact Scale (SIS) versions 3.0, developed by Pamela W. Duncan et al. METHODS: The SIS, a stroke-specific outcome measure, assesses 8 domains: strength (4 items), memory (7 items), emotion (9 items), communication (7 items), activities of daily living/instrumental activities of daily living (ADL/IADL) (10 items), mobility (9 items), hand function (5 items), and participation (8 items). SIS was translated and back-translated according to the standardized guidelines for cross-cultural adaptation. Patients consecutively admitted with ischemic stroke were recruited, until 30 patients in each of three groups of stroke severity were reached. Stroke severity was evaluated with NIH stroke scale and categorized as mild (0-6), moderate (7-15), or severe (16-38). Patients were assessed 3 times, within 7 days of admission, 3 months and 6 months afterwards. Ten randomly selected patients were re-administered 2 weeks after first administration for test-retest study. RESULTS: We studied 50 mild, 36 moderate, and 33 severe stroke patients. Cronbach's alpha was high for 6 domains (0.95 to 0.98), moderate for emotion (0.69) and participation (0.77). Intraclass correlation coefficients ranged from 0.63 to 0.89, except for memory (0.13), hand function (0.47), and ADL/IADL (0.53). Based on first assessment, memory and communication may have potential for ceiling effects in mild stroke group, strength and hand function may have potential for floor effects in moderate stroke group, and all domains may have potential for floor/ceiling effects in severe stroke group. The correlations between each of the physical domains (strength, ADL/IADL, mobility, hand function) and Barthel Index were good (0.72 to 0.92). Memory domain showed a high correlation with MMSE (0.81). CONCLUSIONS: The Mandarin version of the SIS is an acceptable stroke-specific outcome measure in most domains. Further studies in determining the content should enhance confidence in its validity.