Aims/hypothesis Thiazolidinediones (TZDs) are associated with an increased risk of fracture but the mechanism is unclear. We sought to determine the effect of TZDs on bone mineral density (BMD) and bone turnover markers. Methods PubMed, EMBASE and Cochrane CENTRAL databases were searched from inception until January 2015 for randomised controlled trials comparing TZDs with metformin, sulfonylureas or placebo, and those reporting changes in BMD and/or bone turnover markers. The primary outcome was percentage change in BMD from baseline and results were pooled with random effects meta-analyses. Results In all, 18 trials were included in the primary analyses and another two were included in the sensitivity analyses (n=3,743, 50% women, mean age 56 years, median trial duration 48 weeks). TZDs decreased BMD at the lumbar spine (difference −1.1% [95% CI −1.6, −0.7]; p<0.0001), total hip (−1.0% [−1.4, −0.6]; p<0.0001) and forearm (−0.9% [−1.6, −0.3]; p=0.007). There were statistically non-significant decreases in BMD at the femoral neck (−0.7% [−1.4, 0.0]; p=0.06) and total body (−0.3% [−0.5, 0.0]; p=0.08). Five trials (n=450) showed no statistically significant difference in percentage change in BMD between the TZD group and controls up to 1 year following TZD withdrawal. In 14 trials, the effect of TZD treatment on turnover markers varied considerably between individual studies. Conclusions/interpretation Treatment with TZDs results in modest bone loss that may not be reversed 1 year after cessation of treatment.