Purpose: Clinical and observational studies suggest that use of thiazolidinediones (TZDs) is associated with an increased fracture risk. In addition, type 2 diabetes mellitus (T2DM) is a risk factor for osteoporotic fracture. Our aim was to estimate fracture risks in TZD users and users of other antidiabetic drugs, classified according to proxies of disease severity. Methods: We conducted a population-based cohort study utilizing the Dutch PHARMO database (1998-2008). PHARMO links pharmacy-dispensing data to the National Hospital Registry. Oral antidiabetic users (n=123452) were matched 1:4 by year of birth and sex to non-users. Cox proportional hazards models were used to estimate hazard ratios (HRs) of fracture in TZD users. We created a proxy indicator for disease severity. The first stage was defined as current use of either a biguanide or a sulfonylureum, the second stage as current use of a biguanide and a sulfonylureum at the same time, the third stage was assigned to patients using TZDs and the fourth stage to patients using insulin. Results: The risk of osteoporotic fracture was increased 1.5-fold (HR 1.49, 95%CI 1.28-1.73) in patients who currently used TZDs (stage 3), and for patients using insulin (stage 4), the risk was increased 1.2-fold (HR 1.24, 1.14-1.36), as compared with controls. In the first and second stages, risks were lower: HR 1.11 (1.06-1.17) for stage 1 and HR 1.03 (0.96-1.11) for stage 2. Conclusions: When observational studies assess risk of fracture in patients with TZDs, the severity of T2DM should be taken into account. © 2012 John Wiley and Sons, Ltd.