Most medical cost-effectiveness analyses include future costs only for related illnesses, but this approach is controversial. This paper demonstrates that cost-effectiveness analysis is consistent with lifetime utility maximization only if it includes all future medical and non-medical expenditures. Estimates of the magnitude of these future costs suggest that they may substantially alter both the absolute and relative cost-effectiveness of medical interventions, particularly when an intervention increases length of life more than quality of life. In older populations, current methods overstate the cost-effectiveness of interventions which extend life compared to interventions which improve the quality of life.