BACKGROUND: The results of clinical trials have suggested that the thiazolidinedione antidiabetic agents rosiglitazone and pioglitazone are associated with an increased risk of fractures, but such studies had limited power. The increased risk in these trials appeared to be limited to women and mainly involved fractures of the arm, wrist, hand, or foot: risk patterns that could not be readily explained. Our objective was to further investigate the risk of fracture associated with thiazolidinedione use. METHODS AND FINDINGS: The self-controlled case-series design was used to compare rates of fracture during thiazolidinedione exposed and unexposed periods and thus estimate within-person rate ratios. We used anonymised primary care data from the United Kingdom General Practice Research Database (GPRD). All patients aged 40 y or older with a recorded fracture and at least one prescription for a thiazolidinedione were included (n = 1,819). We found a within-person rate ratio of 1.43 (95% confidence interval [CI] 1.25-1.62) for fracture at any site comparing exposed with unexposed periods among patients prescribed any thiazolidinedione. This association was similar in men and women and in patients treated with either rosiglitazone or pioglitazone. The increased risk was also evident at a range of fracture sites, including hip, spine, arm, foot, wrist, or hand. The risk increased with increasing duration of thiazolidinedione exposure: rate ratio 2.00 (95% CI 1.48-2.70) for 4 y or more of exposure. CONCLUSION: Within individuals who experience a fracture, fracture risk is increased during periods of exposure to thiazolidinediones (both rosiglitazone and pioglitazone) compared with unexposed periods. The increased risk is observed in both men and women and at a range of fracture sites. The risk also increases with longer duration of use.