OBJECTIVES: To compare EQ-5D and SF-6D utilities in chronic heart failure (CHF) patients with varying levels of disease severity, proxied by the Duke Activity Status Index (DASI). METHODS: A consecutive sample (N=251) of CHF patients undergoing elective cardiac surgery were surveyed with the EQ-5D, SF-36 and the DASI, with the latter used to perform a uniform 5-way split of the sample according to functional capacity. Association and level of agreement between instruments in each severity group were estimated with Pearson’s r and the intraclass correlation coefficient (ICC) respectively. Paired-samples t-test was used to identify significant score differences, which were regarded as minimally important differences (MID) when exceeding 0.03. In a linear regression model, the DASI score was used as an anchor of disease severity to identify a potential “crossover” point between EQ-5D and SF-6D utilities. RESULTS: EQ-5D and SF-6D were in agreement and strongly correlated over the entire sample (ICC=0.484, P<0.001 and r=0.647, P<0.001), but correlation varied within the DASI-based groups. In the less severe groups (higher functional capacity) EQ-5D scores were significantly higher than SF-6D scores (P<0.001) and differences constituted MIDs. Contrarily, in the more severe groups SF-6D was predominantly higher than EQ-5D. The regression model indicated a utility crossover point at 0.716 and predicted that individuals with a utility score less than this would score higher on the SF-6D than on the EQ-5D, and vice versa. The DASI score at crossover was calculated at 31.99. CONCLUSIONS: In subgroups of patients differing in CHF severity, mean EQ-5D and SF-6D indices differed significantly, implying that the two instruments are not interchangeable and the choice of instrument for measuring health-related utility can lead to variations in cost-utility analyses. Based on evidence provided here, discrepancies in the instruments’ measuring range generate utility differences which require further investigation.