OBJECTIVE: To determine whether women with pelvic endometriosis are at increased fracture risk. DESIGN: Historical cohort study. SETTING: Population-based investigation using the data resources of the Rochester Epidemiology Project. PATIENT(S): From Olmsted County, Minnesota, 987 women with endometriosis that was first surgically visualized or histologically proven in 1970-1989. INTERVENTION(S): None, observational study. MAIN OUTCOME MEASURE(S): Follow-up for fractures through complete inpatient and outpatient community medical records. RESULT(S): In 17,408 person-years of follow-up, 256 women experienced 449 different fractures. The cumulative incidence after 20 years (30.8%) was not elevated relative to that expected (30.6%). The independent predictors of any fracture included age (hazard ratio [HR] per 10 years: 1.61; 95% confidence interval [CI] 1.42-1.84]), corticosteroid use (HR: 2.78; 95% CI 1.48-5.24), prior hip, spine, or forearm fracture (HR: 1.82; 95% CI 1.10-3.02), and use of the selective estrogen receptor modulators, tamoxifen or raloxifene (HR: 4.34; 95% CI 2.14-8.81); physical activity was protective (HR: 0.40; 95% CI 0.18-0.88). There was no significant influence on fracture risk of surgery or other medical treatments for endometriosis. CONCLUSION(S): Despite reported adverse effects of treatment on bone density, there was no overall increase in long-term fracture risk in this unselected cohort of women with proven endometriosis.