A dementia screening survey was carried out in Shanghai using a culturally adapted Chinese version of the Mini-Mental State Examination. A probability sample of 5055 community-dwelling elderly in Shanghai was surveyed, 1497 aged 55-64, 2187 aged 65-74, and 1371 aged 75 and over. In the 73.3% of the subjects who had gone to school, using the age and education adjustments suggested by Kittner et al. (1986), [Kittner et al. J Chron Dis 39: 163-170; 1986] suitable cutoff scores could readily be selected to identify the subjects who should be examined intensively for the presence of dementia. However, in the 26.7% who had not gone to school, there was a significant increase in low scores on the mental status test as well as a different error pattern, reflecting the lack of formal education. Methods for following cognitive changes in illiterate individuals need further development.