The promise of cost-effectiveness analysis in health care is that it allows policymakers to compare what at first blush are apples and oranges. CEA purports to show how much health benefit is likely to be produced by very different investments. It attempts to express the effectiveness of health care interventions and programs in common units of value, such as adjusted life (QALYs), that allow the benefits of various treatments to be quantified on the same scale of measurement.[1] The work CEA can do is indeed remarkable. It can take programs with qualitatively very different outcomes--life-saving dialysis and quality-enhancing hip replacements, for example--and inform us of their costs in relation to their comparable effects--their QALYs. This ability to compare disparate effects is one of several respects in which CEA is the child of utilitarian welfare economics. In that larger discipline too, a huge range of values gets collected into one common and measurable notion, Moreover, welfare economics' conception of to society is typically built up out of individuals' utilities, combined in some way. In all of these respects--its one scale of value, its quantifying of that value, and its combining of individual pieces of value into aggregate wholes for society--conventional health economics is just what one would expect of a specialty within economics. How does CEA in health care obtain its single metric of value for comparing widely different effects? Typically three factors are incorporated. The first, and by far the most conceptually complex, is the of the quality improvement that treatment produces: the saving of life and its maintenance at a certain level. This change is expressed as a savings or improvement of quality of life. Call it the size of treatment effect. It is here that mortality and morbidity are compared and combined. Death is assigned the value 0, full health the value 1.0, and all other health states better than death are arrayed in between, from most to least severe, on the basis of responses that interview subjects give to certain questions. Various types of questions are used to elicit opinions about quality of life. In time trade-off questions, for example, respondents may be asked how many of an anticipated twenty remaining years of their lives they would be willing to sacrifice in order to obtain the complete cure of a specified health condition. Suppose that on average their answer is four--that is, 20 percent of their remaining time. Then they have rated the quality of life in that state of health at 0.8, a 20 percent reduction from 1.0. The second factor used to construct comparable units of value is the duration of that health improvement, and the third factor is the number of persons receiving it. Conventional CEA simply multiplies these factors: the of treatment effect is multiplied by both the duration of that benefit and the number of beneficiaries. For example, if hip replacements generally raise recipients' quality of life from 0.8 to 0.98, effectively last 10 years, and cost $18,000 each, they typically produce 1.8 QALYs at a cost of $10,000 per QALY. If a year of inpatient hemodialysis typically costs $32,000 and saves a life of 0.8 quality, it produces 0.8 QALYs at a cost per QALY of $40,000. If the medical demand for these two procedures were not being met, these numbers suggest that it would be better to invest in additional hip replacements rather than in expanded dialysis. The quantified effects used in CEA are classic expressions of a form of utility--what might be called health-related utility. While CEA is technically a descriptive analysis because it does not actually tell decisionmakers to maximize utilities, it reveals which categories of health care investments will maximize them. Utilitarianism is its philosophical parent. …