OBJECTIVE: To assess the equivalence of English and Chinese versions of the SF-36. METHODS: Using a crossover design with block randomisation and stratification by age, identical English or Chinese questionnaires containing the English (UK) and Chinese (HK) SF-36 versions were administered 3-16 days apart to 168 free-living, bilingual, ethnic Chinese volunteers in Singapore. Item level equivalence of both versions was assessed by comparing item means and orderings within each scale. Scale level equivalence was assessed by comparing internal consistency (Cronbach's alpha), results of factor analysis and mean scale scores (using paired t-tests and intra-class correlations). RESULTS: Item and scale level comparisons supported the equivalence of both versions. For both the versions, item means, item ordering and Cronbach's alpha were similar, and factor analysis yielded two factors with similar factor loadings. There was no clinically important difference in mean scale scores for seven of eight scales, and intra-class correlations were excellent/good for five scales (0.69-0.77) and moderate for three scales (0.55-0.57). CONCLUSION: English (UK) and Chinese (HK) SF-36 versions are equivalent in bilingual Singapore Chinese. Our data suggest that SF-36 scores from English- and Chinese-speaking subjects may be combined in studies using the SF-36, increasing the power and representativeness of such studies.