Evidence that the productivity of marginal expenditures on health care is low is reviewed. A number of reasons why expenditure on health care continues to be high in the light of such evidence are presented. Priorities for research in health economics are then set out.