This study reports the methods and initial findings of the first longitudinal study of Alzheimer's disease and dementia in China. A probability sample of 5,055 noninstitutionalized elderly persons in Shanghai was tested directly during the first phase of the study using a Chinese version of the Mini-Mental State Examination (MMSE). Details of sampling design and data collection procedures are described. Overall, some 4.1 percent of adults 55 years or older may be classified as having severe cognitive impairment, and 14.4 percent are mild cases. The rates for females are higher than for males by a ratio of 3.75 in the severe category, and 2.6 in the mild group. Within each age group, cognitive impairment rates vary by education. Multiple logistic regression was used to examine, within each age group, the nature of the association between the presence of a cognitive impairment and educational level controlling for sex. The results showed that educational attainment has a highly significant inverse relationship with prevalence of cognitive impairments (severe vs others). On the other hand, when educational attainment was controlled for in the logistic regression model, sex was significantly associated with prevalence of cognitive disorders for the age groups 65-74 and 75 years or older, but not for the group 55-64 years. These findings suggest the impact that basic educational deficits have on human cognitive functioning as measured through tests like the MMSE. Cross-tabulations of impairment rates according to marital status, economic status, and health-related problems are also presented.