Cost of illness (COI) studies describe the economic burden of disease on society. In this article a standard procedure for a COI study is developed, including the explicit definition of the disease, choice of relevant variables and appraisal of direct and indirect costs. COI studies can be incidence-based or prevalence-based. The adjustment of cost figures for time preferences and the performance of a sensitivity analysis are presented. The standard methodology is applied to diseases in two different areas. The first disease category is dyspepsia, a complaint with a rather somatic background. The second is schizophrenia, a mental syndrome. In performing COI studies in practice, however, researchers are forced to deviate, in many aspects, from the theoretical standards. In this article these choices, and the reasons behind these choices, are explained. Furthermore, we discuss certain problems regarding the reluctance to make a diagnosis regarding certain diseases, the reliability and the validity of the sources used and the absence of certain figures. The value of the information derived from COI studies for policy-making is assessed.