Indirect costs of disease often constitute a substantial part of estimated costs or savings in economic evaluations of healthcare programmes. The human capital approach is almost unanimously used for estimating indirect costs, defined as production loss due to disease, although a growing number of authors question its validity. This article discusses the relevance of indirect cost estimates for health policy and reviews the current empirical and methodological literature on this issue. It describes several important issues and controversies regarding indirect costs, such as the consequences of short term absence from work for productivity, reduced productivity without absence from work, the influence of unemployment on production loss, the relation between health effects and indirect costs, and the possible medium term macroeconomic consequences of absence from work and disability. It concludes that indirect costs are relevant for health policy, provided that the estimates of indirect costs reflect the real changes in production due to disease, including the production of unpaid labour. Future research should focus on attaining these estimates. Indirect costs in economic evaluations should preferably be presented separately from direct costs, health effects and other study outcomes.