OBJECTIVES: Compare and evaluate the conceptual and numerical differences between standard gamble (SG) and conjoint analysis (CA) utility weights. METHODS: Reducing complex relationships among health outcomes to a single utility index requires strong assumptions. We compare and evaluate the required assumptions for SG and CA utility weights using numerical estimates from published SG weights and CA utility estimates from a study of acute respiratory and cardio-vascular symptoms. We analyze alternative mappings of one index into the other, evaluate necessary assumptions and restrictions, and compare empirical differences. RESULTS: Differences in the underlying metrics and information content of SG and CA utility indexes complicate mapping one scale into the other. Even when the scales can be defined comparably for a given range of outcomes, extrapolating CA utility outside this range may yield corresponding SG weights outside the 0-1 interval. CA methods facilitate including other utility-relevant health-care features, such as dosing frequency, employ cognitive tasks more familiar to patients than the SG task, and can be scaled in natural units such as money or time. However, CA utility can vary nonlinearly by duration, health-care process attributes, and individual characteristics and thus cannot generally be summed across individuals, health states, and time periods. If decision makers require a simple, aggregate utility measure, they must accept SG restrictions that break the correspondence between measured utility and actual patient preferences. CONCLUSIONS: CA utility weights have several conceptual and empirical advantages relative to SG utility weights. The perceived usefulness and practicality of SG utility for constructing QALYs must be weighed against the potential for providing decision makers with misleading information about the net benefits of health interventions.