Objective: Malnutrition and low weight in eating disorders (EDs) are associated with increased fracture risk compared to the general population. In a cohort study, we aimed to determine fracture rates compared to age and gender matched controls (ratio 5:1), assess the impact of disease remission on fracture risk, and establish predictive factors for fractures. Method: Of note, 803 ED patients referred to specialized ED treatment between 1994 and 2004 were included. In 2016, data on fractures were obtained through the Danish National Registry of Patients. Results: Fracture risk was increased in anorexia nervosa (AN; IRR 2.2 [CI 99%: 1.6–3.0]) but not in bulimia nervosa (BN; IRR 1.3, ns) or other specified feeding or eating disorders (OSFED; IRR 1.8, ns). IRR in the AN group were increased for vertebral fractures (IRR 3.8 [CI 99%: 1.4–10.3]), upper arm (IRR 3.0 (CI 99% 1.6–5.5) and hip (IRR 6.6 [CI 99%: 2.6–18.0]). Disease remission in AN is associated to lower fracture risk compared to active disease, but higher fracture risk compared to controls (IRR 1.7 [CI 99%: 1.1–2.7]). In regression analysis, age at debut of disease, nadir BMI and duration of disease before referral to treatment, independently predicted fracture. Discussion: We confirm increased fracture risk in AN, and show significant differences in fracture risk between patients in disease remission and patients with active disease. Furthermore, we show that age at debut of disease and duration of disease before referral to treatment is positively correlated to fracture risk, whereas nadir BMI is negatively correlated to fracture risk.