OBJECTIVE: To predict utility values from health-related quality of life (HRQL) assessments could, if possible, be a fruitful way to use HRQL assessments for health economic evaluations. With this in mind, we compared values for SF-6D, an index derived from the domain values of SF-36 by an algorithm from Brazier et al., with values from the EuroQol instrument, the EQ-5D index, and the EQ-VAS (Visual Analog Scale). METHODS: Data from two surveys of patients with respiratory disease were used in this comparison. The first data set was from 120 patients in the northern part of Sweden with COPD, and the other dataset from 206 patients in Hungary suffering from asthma. Both surveys covered patients with different severity of the disease. The two datasets were analyzed separately. The EQ-VAS values were rescaled from a range of 0-100 to a 0-1 range. RESULTS: Results are consistent for the two different data sets and shows an expected gradient for severity groups in both cases. SF-6D has on the average slightly lower values than EQ-5D but higher than EQ-VAS (Mean for COPD: 0.74 vs 0.78 and 0.65; ASTHMA: 0.69 vs 0.70 and 0.62). On the other hand, both EQ-5D and EQ-VAS have larger variation with a substantial proportion of patients reaching maximum or minimum achievable values. Correlations between the different indices are deceptively high because of extreme values (r for COPD: 0.75-0.69; ASTHMA: 0.70-0.49. CONCLUSION: SF-6D shows a response pattern more in line with intuitive expectation than EQ-5D and EQ-VAS, which both seem to have a ceiling (and floor) problem.