OBJECTIVES: When using utility weights in economic evaluation, it is standard to assume the trade-offs individuals are willing to make between quality of life and length of life are constant. However, this may not be reasonable, particularly when comparing preferences under relatively long or short life expectancies. A number of patterns that differ from the standardly assumed pattern are possible, and the objective of this work is to test a range of these, particularly proportionality, the degree of non-linearity of utility with respect to time, and the scale of the utility function when anchored to the 0-1 scale required for construction of the quality-adjusted life year (QALY). METHODS: A discrete choice experiment was administered in an online sample of 981 Australians. The sample was selected to be age- and gender-representative.  Each respondent saw 16 choice pairs, 8 with shorter durations (one of 2,4,8 or 16 months) and 8 with longer durations (one of 2,4,8 or 16 years). Data for each set of eight were analysed separately using a range of regression approaches which considered respondent heterogeneity. RESULTS: Trade-offs between quality of life and life expectancy, and between different aspects of quality of life, do not systematically differ based on the range of life expectancies considered. Therefore, the scales of the utility algorithms resulting from either subset of data do not appear to substantially differ. Non-linearity of utility with respect to time is apparent both when life expectancies are relatively short or long. CONCLUSIONS: Health state valuation algorithms do not appear to be strongly affected by the selection of the range of life expectancies. While health state valuation tasks should consider a range of durations, the impact of not doing so appears to be modest.