Background The aim of the study was to determine the construct validity of the 12‐item Short Form health survey questionnaire (SF‐12) across ethnic groups in a large community sample of the United Kingdom.

Methods A postal survey was carried out in English using a questionnaire booklet, containing the SF‐12 and a number of other items relating to experiences of chronic illness and utilization of health care services. The dataset was the National Survey of NHS Patients. The sample consisted of 1000 residents within each Health Authority in England who were randomly selected from the electoral registers, giving an initial sample of 100 000.

Results A total of 61 426 (61.4 per cent) questionnaires were returned; 94.3 per cent of respondents classified themselves as white and 5.7 per cent classified themselves as members of other ethnic groups. Construct validity of the SF‐12 was assessed by comparing results from the two summary scores (the Mental Health Component Summary (MCS) score and the Physical Health Component Summary (PCS) score) with overall self‐assessed health and limiting longstanding illness. Although there were generally consistent patterns of association between overall self‐assessed health or limiting longstanding illness and the MCS and PCS scores in all the ethnic groups, there were significant differences between the MCS and PCS scores of Indians, Pakistanis and Bangladeshis who understood English fluently and those who did not. Furthermore, there were differences in the completion rates of the SF‐12 between ethnic groups and a reversal of the general pattern of increasing MCS scores with increasing age in Bangladeshis.

Conclusion The results indicate that the use of the SF‐12 to measure the health of ethnic minorities seems acceptable in most instances, but may prove problematic in those instances where respondents complete the questionnaire via an untrained translator, such as a friend or family member. The systematic differences in MCS and PCS scores between ethnic minorities who understood English fluently and those who did not suggest that the meaning of specific SF‐12 items may change when informally translated. Future research using the SF‐12 to measure the health status of ethnic minorities in the United Kingdom via postal surveys must include questions on whether respondents completed the questionnaires via informal translations. In general, those wishing to measure the health of members of ethnic groups who are unable to read English might consider using different techniques to gain the information from these groups.