Objectives: Expressing the outcomes of treatment in quality-adjusted life years is increasingly important as a tool to aid decision makers concerning the allocation of scarce resources within the health care sector. A quality-adjusted life year is a measure of life expectancy that is weighted by health-related quality of life. These weights are referred to as utility scores and are usually measured by multiattribute utility measures. Several studies found that different utility measures provide different estimates of the same person's level of utility. The aim of this study was to investigate which of two widely used utility measures, the EQ-5D and the HUI mark III, is preferred in a tinnitus population. Methods: Baseline and follow-up data on EQ-5D and HUI mark III of 429 patients of a randomized controlled clinical trial, investigating cost-effectiveness of usual care versus specialized care of tinnitus, were included. Agreement, discriminative power, and responsiveness of the health state description and the utility scores were examined. Results: Corresponding dimensions of the EQ-5D and HUI mark III showed large correlations; although ceiling effects were more frequently observed in the EQ-5D. Mean utility scores for EQ-5D (0.77; SD 0.22) and HUI mark III (0.64; SD 0.28) were significantly different (Wilcoxon signed ranks test, p <0.001), and agreement was low to moderate (intraclass correlation coefficient = 0.53). Both health state description and utility scores of both measures discriminated between different severity groups. These groups were based on baseline scores of the Tinnitus Questionnaire. The HUI mark III had a higher ability than the EQ-5D to detect improved patients from randomly selected pairs of improved and unimproved patients. Conclusion: This study shows that different utility measures lead to different health state descriptions and utility scores among tinnitus patients. However, both measures are capable of discriminating between clinically different groups. The HUI mark III is more responsive than the EQ-5D, and therefore preferred in a tinnitus population.