There remains considerable disagreement concerning the preferred generic utility-based measure of health-related quality of life for use in constructing quality-adjusted life years. The recent appearance (in a published form) of a new measure, the SF-6D, has highlighted this issue. The SF-6D and EQ-5D have many similarities, but marked variation has been shown in the results generated by the two instruments. The study reported here is an exploration of why such divergent results exist. There are two possible explanations: variation in the descriptive component of the instruments and variation in the values applied to health states. The results suggest two important conclusions. First, the SF-6D can describe severe health states, including states that (according to the EQ-5D scoring algorithm) are viewed as worse than the state of being 'dead'. Second, much of the large discrepancy between the results generated using the two instruments appears to stem from very different valuations being placed on similar health states.