BACKGROUND: A number of different measures can be used within cost-utility analyses, we compared results according to both the EQ-5D and SF-6D. METHODS: A randomized trial was conducted to compare 4 options for people with knee pain. Over the 2 year trial period, the change in cost to health-service was estimated, and both the EQ-5D and SF-6D were used to estimate the change in quality-adjusted life years (QALYs). Using a complete case analysis, the cost-utility (incremental cost-effectiveness ratio [ICER]) of each option, according to both the EQ-5D and SF-6D, was calculated and assessed in relation to the cost-effectiveness threshold of pound20,000 per QALY. RESULTS: Of the 389 participants, 247 had complete cost, EQ-5D and SF-6D data. According to the EQ-5D, option 1 had an estimated ICER of pound10,815 (compared with option 4), option 2 was dominated by option 1, and option 3 was subject to extended dominance. Conversely, according to the SF-6D, option 3 had an ICER of pound9999 (compared with option 4), option 2 had an ICER of pound36,883 (compared with option 3), and option 1 was subject to extended dominance. CONCLUSION: The EQ-5D and SF-6D estimated that different options (1 and 3, respectively) were cost-effective at the pound20,000 per QALY threshold, demonstrating that the choice of measure does matter.