We investigated the effects of galantamine on cognitive subdomains in Alzheimer's disease (AD). Sixty-six patients with mild-to-moderate AD received open-label galantamine for 52 weeks. Cognitive function was measured using the Korean version of the Alzheimer's Disease Assessment Scale-cognitive subscale (ADAS-cog-K). Response to galantamine was defined as "improvement or no deterioration" on the total scores of the ADAS-cog-K at 26 weeks. In the overall intent-to-treat sample, we found less cognitive decline during 26 and 52 weeks than the expected untreated course as predicted by Stern's equation. The operationally defined response rate at 26 weeks was 66.7%. The responders differed significantly from the nonresponders only in the memory and language domains but not in the domains of praxis or frontal/executive function or in secondary outcome measures of neuropsychiatric symptoms and activities of daily living. The subdomain analysis revealed an effect of galantamine on preservation of memory that was not apparent in the overall analysis. Failure to achieve responder status by 26 weeks was associated with no further possibility of response.