SUMMARY The Mini-Mental State Examination was translated into Cantonese and administered to 111 normal elderly and 79 demented patients. The scale was found to have good reliability and validity as an instrument to detect cognitive impairment in our locality. A cut-off score of 19/20 is recommended as indication for further evaluation of cognitive impairment. Keywords: Cantonese, Mini-Mental State Examination, cognitive impairment INTRODUCTION The Mini-Mental State Examination (MMSE, Folstein et al, 1975) is a widely used instrument for assessing cognitive functions both in clinical settings and in research. The MMSE is divided into two sections, the first of which requires oral responses only and covers, orientation, memory and attention. The second part tests subjects' ability to name objects, follow verbal and written commands, write a sentence and copy a complex polygon. The maximum total score is 30 and a cut-off point of 23/24 was reported to be able to discriminate between patients with cognitive impairment and normal subjects (Folstein et al, 1985). It has been validated in various studies conducted in Western countries. However, recent studies have found that the specificity of the MMSE was lower for individuals with less than 8 years of schooling and for those over the age of 65 (Anthony et al, 1982; Holzer et al, 1983). Furthermore, cultural differences can also influence the scores and performance on the MMSE (Escobar et al, 1986; Salmon et al, 1989). Recently, several Chinese versions of the MMSE have been developed and used in surveys for dementia in China (Li et al, 1989; Yu et al, 1989). The version used in Shanghai (Yu et al, 1989) is of particular interest as it was carefully adapted from the original MMSE and has been used in a large scale study involving 5,055 elderly persons in Shanghai. However, this is a Mandarin version which has not been used in our locality. As the elderly population is rising rapidly in Hong Kong, from less than 5% 10 years ago to 8.7% in 1991 for people at or above 65 years of age (1991 Census), there is an increasing need to develop screening instruments for the assessment of cognitive impairment for clinical and service planning purposes. An epidemiological survey of cognitive impairment in the elderly has been carried out in Hong Kong recently (Liu et al, 1993), using a Cantonese version of MMSE which has not been validated in Hong Kong. Fan (1992) reported a validation study of another Cantonese version of the MMSE based on the findings of a small sample of psychogeriatric outpatients. However, this study suffers from a number of limitations and they include: 1) the small number of subjects (N=29), 2) the failure to provide a control sample of normal subjects for comparison purpose and 3) the lack of information on the psychometric properties of the scale such as internal consistency and test retest reliability. In this study, the MMSE was translated into Cantonese and the scale's reliability and validity were examined, using a sample of Chinese elderly in Hong Kong. METHOD THE INSTRUMENT The MMSE was translated into Cantonese and back-translated by a team of bilingual psychiatrists. Most of the items on the MMSE could be directly translated and used in Hong Kong. Major adaptations, however, were necessary for several items. The phrase "No ifs, and's, or but's" was changed to a Cantonese phrase of 'Uncle buys fish intestine [TEXT NOT REPRODUCIBLE IN ASCII.]" which is an alliteration in Cantonese. The phrase "Please close your eyes" was replaced by the phrase "Clap your hands" as the original phrase may sometimes be interpreted to be related to death in the Chinese culture. For the writing test, we adopted the strategy used in the Chinese MMSE (Katzman et al, 1988), i.e. the respondent was asked to "Say a sentence" instead of writing a sentence, to avoid failure on this item as a result of lack of education. …