This article reviews the cost of illness of ankylosing spondylitis in the literature and identifies limitations of comparability of cost of illness studies. The literature was searched semisystematically for studies that aimed to assess the cost of illness of ankylosing spondylitis from the societal perspective. Studies were appraised for methods and results following a self-composed checklist. To compare the aggregated costs between the studies, adjustments for differential timing and purchasing power parities between the countries were applied. In total, 53 titles were retrieved by a MEDLINE search. Five articles reported on the costs of illness in four patient populations, one from the USA and three from Europe. All studies were prevalence studies with a bottom-up approach and reported direct and productivity costs. Patient characteristics differed with respect to sampling source, age, disease duration, presence of spondylitis-related comorbidity and employment status. Categories of resource use and costs were especially difficult to compare with regard to visits to types of healthcare providers and use of formal and informal help in relation to inability to perform unpaid work. In addition, not all studies reported the productivity costs based on human capital as well as the friction cost method. The cost per unit of resource use was only provided explicitly in one publication. Sensitivity analyses were usually not performed. Total 2002 costs based on the human capital approach varied between USD 7243 and 11,840, and productivity costs accounted for 53-73% of the total costs. Total 2002 costs based on the friction costs varied between USD 3353 and 3903, and productivity costs accounted for 15-26% of the total costs. Cost drivers of the direct costs varied among the studies. Physical functioning and/or disease activity were consistent determinants of total costs. In conclusion, between four bottom-up prevalence studies in ankylosing spondylitis from different countries, there were differences in characteristics of patients studied, inclusion of unpaid help and unpaid production loss in the costs and choice of human capital compared with the friction cost method to calculate the productivity costs. The absence of information of the unit-cost per resource hampers comparability. Overall, the ankylosing spondylitis-related costs of illness are substantial and the high costs of formal or informal help and work disability reflect the impact of the disease on physical functioning.