ObjectivePreclinical studies suggest that thiazolidinediones (TZDs) may have a protective effect on rheumatoid arthritis (RA), but evidence from population‐based studies is scarce. This study aimed to assess the association between use of thiazolidinediones and incidence of RA in a retrospective cohort of patients with type 2 diabetes mellitus (T2DM).MethodsA retrospective cohort of T2DM patients who were new users of TZDs or alpha glucosidase inhibitors (AGIs) was assembled. We applied the inverse probability of treatment weighted (IPTW) Cox model to estimate the hazard ratio (HR) of RA incidence associated with the use of TZDs compared with AGIs.ResultsThe final analysis included 56796 new users of AGIs and 14892 new users of TZDs. The incidence of RA was 187.4 and 135.2 per 100,000 person‐years in AGI users and TZD users. Compared with use of AGIs, TZD use was associated with a reduction in RA incidence, with an HR of 0.72 (95% CI, 0.59–0.89). HRs for cumulative use of TZDs for 0.51–4.0 years and > 4 years with incidence of RA, were 0.55 (95% CI, 0.35–0.88) and 0.74 (95% CI,0.57–0.98), respectively. Various subgroup analyses and sensitivity analyses were consistent with the primary analysis.ConclusionUse of TZDs is associated with a decreased risk of incident RA in T2DM patients.