A key recommendation of the recent Panel on Cost-Effectiveness in Health and Medicine was that cost-effectiveness analyses be carried out from a societal perspective. The authors show that two of the Panel's recommendations concerning costs are not consistent with a societal perspective, and how to correct those inconsistencies. In its recommendations concerning costs resulting from morbidity, the Panel advises excluding lost income from costs in the belief that individuals take income changes into account when they respond to the quality-of-life questions that are used to calculate quality-adjusted life years (QALYs). It is shown that even if individuals do consider income changes in responding to these quality-of-life questions, this recommendation would seriously underestimate production losses due to morbidity, since individuals do not bear a major part of lost production. In its recommendations concerning costs resulting from mortality, the Panel does not require that health care costs for "unrelated" illness and non-health care consumption and production during added life years be included in the Reference Case. It is shown that omitting these costs will seriously distort comparisons of programs at different ages and favor programs that extend life over those that improve quality of life. This can be corrected by including total consumption minus production in added life-years among costs.