Retrospective claims analysis indicated that high levels of daily and cumulative doses of systemic glucocorticoids were associated with elevated fracture risk in a large cohort of new RA patients under age 65. Heightened risk began to decline within months of discontinuation. Findings were similar among patients age <50 years. INTRODUCTION: We evaluated the impact of systemic glucocorticoid exposure on fracture risk among relatively young patients with new-onset rheumatoid arthritis (RA). METHODS: Using administrative data, we identified 42,127 RA patients diagnosed January 1, 2005-December 31, 2012, age 18-64 years, with benefits coverage for ≥12 months before RA diagnosis. Follow-up extended to clinical fracture, cancer diagnosis, or December 31, 2012. Glucocorticoid users were new to therapy. Fracture incidence rates (IR) were stratified by glucocorticoid exposure expressed as prednisone equivalent doses. Cox's proportional hazards models estimated fracture risk adjusted for demographics and baseline clinical characteristics to assess dose-response relationships with current (daily) and prior (cumulative) dose, and by time since discontinuation. RESULTS: Most patients (85 %) had glucocorticoid exposure. Exposed and unexposed patients were demographically similar (74 % female; mean age 49.7 and 48.8 years); 1 % had prior fracture. Fracture IRs (95 % confidence intervals) were 5 to 9 per 1000 person-years at doses <15 mg/day, 16.0 (11.0, 22.6) at doses ≥15 mg/day, and 13.4 (10.7, 16.7) at cumulative doses ≥5400 mg. Adjusted fracture risk was approximately 2-fold higher at highest dose levels compared with 0 mg/day current daily dose and <675 mg cumulative dose, respectively. Fracture risk was 29 % lower at 60-182 days post-discontinuation compared with ongoing use and was similar to unexposed patients by 12 months. Findings were similar among patients age <50 years. CONCLUSIONS: Among younger, new-onset RA patients, fracture risk was significantly elevated at high levels of daily and cumulative dose, and was similar to unexposed patients by 12 months post-discontinuation.