This paper describes the psychometric analysis of the alcohol use disorders identification test (AUDIT) after it was modified for use in Hong Kong and administered to examine the patterns of hazardous and harmful drinking. The modified version of AUDIT was an 18-item instrument in which 10 items were completely adopted from the original version and 8 items were added to improve its cultural sensitivity. It was translated into Chinese and back translation was undertaken to confirm the equivalence of the Chinese and English versions. Following a pilot study the instrument was administered to 450 subjects who were recruited from two acute general hospitals, a University Health Clinic and three community health centres. The content validity was judged as adequate by a panel of five international and local experts and the instrument achieved a high reliability coefficient of 0.99 during a test retest procedure conducted with 20 subjects. Factor analysis was performed on the responses obtained from 450 subjects which supported the construct validity of the 18-item instrument. The modified instrument had a consistently high internal consistency reliability (Cronbach's alpha = 0.96-0.97) when tested in the different settings. It was found that a higher percentage of respondents from the hospitals (14.5%) drank at a hazardous or harmful levels compared to those from the community (6.2%) or the University (5.3%). The AUDIT proved a reliable and valid measure with potential applications in Chinese cultures. Early intervention and identification of 'at risk' drinking by the AUDIT is supported as a strategy to be implemented by nurses in primary and secondary health care settings in Hong Kong, where there are indications of increasing alcohol overuse.