The objective of this study was to validate the SF-12 Health Survey in heart and stroke patients using a community-based study. Between November 1995 and August 1996, 3,362 patients were invited to join the Hunter Heart and Stroke Register in New South Wales, Australia and to complete the SF-12 Health Survey. Of the 3,362 patients, 2,341 (70%) returned the SF-12. Of those 2,341 patients, 78% completed all 12 items. Those who did not complete the questionnaire were significantly more likely to be females, older, less educated, have stayed longer in hospital and been admitted on emergency. The SF-12 demonstrated construct validity in an analysis restricted to the 1,831 patients who completed the questionnaire: scores measuring physical and mental health status were statistically significantly higher in men than women, in younger than older, in those who had shorter than longer lengths of stay in hospital, in patients whose hospital admissions were planned than emergencies and in heart than stroke patients. Construct validity of the SF-12 among patients able to complete the SF-12 suggests considerable potential for its use in assessing health status in large-scale surveys. However, caution should be taken with the heart and stroke population because of a relatively high in completion rate.