OBJECTIVES: The SF-6D and the EQ-5D are two ways to generate utility values. Many studies compared these methods, but there is limited information in schizophrenia. The objective of this study is to compare the SF-36 and EQ-5D utility values and to compare their evolution over time in European schizophrenic patients. METHODS: We used data from EuroSC, a multicenter 2-year cohort study conducted in France, England and Germany. In several subpopulations, we used paired-samples t-test to identify significant differences at baseline, and calculated the Pearson correlation. Changes from baseline were also calculated at 6 and 24 months. RESULTS: The overall sample was composed of 1088 patients. Mean utility scores were significantly different at baseline (EQ-5D: 0.73 (0.28); SF-6D: 0.65 (0.09); p<0.0001), and values modestly correlated (r = 0.314, p<0.001). The SF-6D scores distribution was found to be normal, contrary to EQ-5D, highly negatively skewed. Range of values was much larger for EQ-5D (EQ-5D: [-0.35; 1.00]; SF-6D: [0.32; 0.87]). The difference was also significant at baseline when considering subpopulations, with higher values for EQ-5D. When considering patients whose illness was stable or had improved at 6 months, change from baseline was significantly different from 0 for EQ-5D (mean 0.067 (0.29) p<0.0001), but not for SF-6D (0.007 (0.11) p=0.28). The result was not significant for patients whose illness worsened (EQ-5D: -0.040 (0.29) p=0.19; SF-6D: -0.007 (0.12) p=0.56). The same trends were observed after 24 months, using depression scores to evaluate improvement or worsening, and when adjusted on the baseline value. CONCLUSIONS: In our study, EQ-5D tends to generate wider and higher scores in schizophrenic patients. EQ-5D seems to be more sensitive to change than SF-6D. The method by which QALYs have been computed is of importance in an economic evaluation.