Economic evaluations of interventions to lower blood pressure or cholesterol have used different outcome measures, or end points, in the denominator. Some have related the costs of interventions to improvements in physiologic end points such as mm Hg reduction in blood pressure. Some have related costs to avoidance of coronary heart disease (CHD) events or gains in life expectancy. Others have measured improvements in outcome in quality-adjusted life years (QALYs) gained. The different end points imply different analytic perspectives and different data requirements. The more ambitious analyses, though potentially more relevant in certain situations, require more controversial assumptions to be made. This paper illustrates the trade-offs of relevance, accuracy, and precision by reference to an evaluation of drug therapy for hypercholesterolemia undertaken in the United Kingdom. Estimates are given of cost per percentage cholesterol reduction, cost per CHD event avoided, cost per CHD-free year gained, cost per life year gained, and cost per quality-adjusted life year gained. In each case the assumptions required and the potential relevance of the estimate are discussed. The main findings are that: 1) some end points cannot be discounted to present values in a meaningful way and hence the timing of costs and outcomes cannot be reflected in the analysis; 2) the incorporation of quality-of-life adjustments for years on drug therapy and years post-CHD events greatly changes the cost-effectiveness ratios; 3) the rate of discount changes the pretreatment level of cholesterol for which cost per life year gained is equivalent to cost per quality-adjusted life year gained.(ABSTRACT TRUNCATED AT 250 WORDS)