We examine five approaches economists and health policy analysts have developed for evaluating policy affecting health a safety: cost-of-illness, willingness-to-pay, cost-effectiveness analysis, risk-risk analysis, and health-health analysis. We examine the theoretical basis and empirical application of each approach and investigate the influences that assumptions embedded in each approach have on policy guidance. We reach four principal conclusions. First, the approaches are not interchangeable: they measure different things. Even estimates using the same approach are often not comparable because, in practice, there is little consistency in the application of any of the approaches. Second, the usefulness of each approach depends on the unit of account. The philosophical decision to eschew the monetization of health costs or benefits constrains the ability of the approach to rank policy options and to gauge the social desirability of policy. Third, all of the approaches except risk-risk analysis and one variation of cost-effectiveness analysis incorporate the effects of income and circumstance. As a result, policy guidance could be influenced by the distribution of income. Fourth, the theory and practice of willingness-to-pay estimation are in opposition. While it is now common practice for regulatory agencies to adopt the willingness-to-pay approach for estimating health and safety benefits, they do so by assuming away the importance of individual preferences. We build on these four conclusions to suggest the appropriate use of each approach.