Several cost-of-illness (COI) studies related to diabetes mellitus have been performed over the last three decades. This review examines the results of these COI studies, identifies the strengths and limitations of the various methods utilised, and suggests future research that will help determine the economic burden of diabetes more accurately. Diabetes imposes a large economic burden on society. The economic cost of diabetes is estimated to be as much as dollars US 100 billion per year in the US alone (1997 values). This estimated cost has increased notably over time, primarily due to price inflation and the increasing prevalence of diabetes. Differing methodologies have significantly influenced the cost estimates and made comparisons between COI studies problematic. For example, early reports tended to rely exclusively on data where diabetes was listed as the primary diagnosis or reason for healthcare use. To better capture the costs associated with diabetes-related complications, later studies have included costs related to diabetes as a secondary or tertiary diagnosis using the attributable risk methodology. Given the types of long-term complications that are associated with diabetes, attempts at capturing these secondary costs are appropriate. However, estimates of attributable risk can be limited by the epidemiological data currently available. The tremendous economic burden of diabetes makes the disease an important clinical and public health problem. In order to formulate an effective response to this problem, it is important to track future economic trends as healthcare delivery, morbidity and mortality patterns evolve. Future research efforts should focus on refining methods to estimate costs, improving the interpretation of study findings, and facilitating comparisons between studies.