This paper examines the performance of the SF-36 as a self-reported outcome measure in a diverse sample of Australian hospital in-patients. The data derive from the Care Continuum and Health Outcomes Project with a total of 2088 randomly selected patients, yielding a response rate of 80%. Distributions, completion rates and, in particular, correlates of measures were used to investigate the validity of the SF-36 according to clinical and psychometric criteria. Physical functioning, bodily pain, role limitations-physical, general health and vitality scales better represented physical than mental health, with the relative strength decreasing in that order; while mental health, role limitations-emotional and social functioning scales better represented mental health, with the relative strength decreasing in order. A cultural dimension was a strong independent correlate of all scales. While the SF-36 performed satisfactorily, there were weaknesses: the social functioning scale was too narrow to cover social health, both role limitations scales had crude response categories and particular subgroups, the frail elderly and those with complex health conditions, required measures with finer gradings for scales such as physical functioning and bodily pain. Further testing is needed to develop self-reports for use in hospital outcomes measurement which is already proposed in Australia.