The article discusses the proposal of some health economists to use the "cost per QALY (quality-adjusted-life year)" ratio as an universal indicator for economic assessment of medical interventions, in the so-called "cost-utility" analyses. Authors argue that QALYs are not a straightforward application of expected utility theory, which is the standard economic model of individual behaviours toward risk and uncertainty. Indeed, QALYs are compatible with economic utility theory only if individuals' preferences regarding health states satisfy certain very restrictive properties: utility independence between length of life and quality of life, constancy of the proportional trade-off between quality of life and length of life, risk neutrality towards health states, constancy through time of the utility associated with each health state. Aggregation of individual QALYs to obtain an indicator for patient groups at the societal level also raises complex equity problems. Last but not least, from the epistemological point of view, QALYs are based on the hypothesis that health interventions only affect the health of the individual and not any other aspects of his well-being. The authors conclude that the "cost per QALY" approach should be abandoned in order to avoid ambiguities that could impede the development of health economics in the medical field.