OBJECTIVES: Health-related quality of life (HRQL) is an essential part of health technology assessment; without this Quality Adjusted Life Years cannot be calculated. The objective of this study was to compare utilities calculated for patients with advanced melanoma in the Phase III clinical trial for ipilimumab (MDX010-20) with utilities produced by vignettes for advanced melanoma valued by the general population. A comparison was also made between standard ‘progression-based’ utilities and those based on the time elapsed between the utility measurement and the patient’s death. METHODS: Utilities from the trial were generated using the EORTC-8D and SF-6D preference-based measures. Analyses by progression status and time to death were conducted on patient-level data, and the prognostic value of the methods was assessed. Patient-level results were then compared with the utilities derived for progressive and non-progressive disease in a separate vignette study. RESULTS: SF-6D and EORTC-8D showed a substantial decrease in utility in the 180 days before the patient’s death (from 0.83 to 0.63 and from 0.66 to 0.51, respectively), which is not consistent with the use of standard Markov progression-based, health-state modelling. Time to death showed a lower Root Mean Squared Error and higher R2when used to predict patient utility, demonstrating that they produce a more accurate assessment of HRQL. Utilities taken from vignettes showed a larger decrease on disease progression (from 0.77 to 0.59) than either the generic SF-6D (from 0.64 to 0.62) or condition-specific EORTC-8D (from 0.80 to 0.76). CONCLUSIONS: Although most oncology modelling is based around disease progression, this may not always be appropriate because the time to a patient’s death appears to be a more accurate predictor of HRQL. This has implications for the analysis of utility information in future cost-effectiveness studies as well as the modelling methods used for oncology treatments and health technology assessments.