Cost-effectiveness analysis can be useful as an aid to decision makers concerned with the allocation of health care resources. The approach requires an explicit, quantitative measure of the health effectiveness of the intervention under analysis, as well as a measure of its net resource cost. Cost-effectiveness analyses are less useful if the measure of effectiveness is expressed in units that are unique to the intervention or class of interventions being considered than if the measure is comparable across interventions. The life year as a unit offers the advantage of comparability across programs, but its use in cost-effectiveness analysis can be misleading if the effects of the program in question include changes in the quality of life. Cost-effectiveness analyses of estrogen therapy in the menopause, high blood pressure control, and coronary artery bypass surgery are used to illustrate the method of analysis and the sensitivity of conclusions to the manner in which quality-of-life impacts are included explicitly (or excluded entirely). Cost-effectiveness analyses, therefore, should always include sensitivity analyses in which preference weights and parameters are varied over the plausible range. Inability to measure the quality of life and preferences regarding such effects should not be an excuse for failing to include them in a cost-effectiveness analysis.