Objective The increased prevalence of osteoarthritis (OA) in postmenopausal women suggests that changes in either circulating sex steroid concentrations or the tissue response to sex steroids may have a role in the pathogenesis of OA. The aim of this study was to examine whether circulating sex steroid concentrations are associated with the incidence of total knee and total hip replacement for OA. Methods Study subjects (n = 2,621; all women) were recruited in 1990-1994 from the Melbourne Collaborative Cohort Study (MCCS). Circulating sex steroid concentrations were measured in blood samples obtained from the women at the time of recruitment. The incidence of total knee and total hip replacement for OA during 2001-2011 was determined by linking the MCCS records to the Australian Orthopaedic Association National Joint Replacement Registry. Results During the followup period, 115 women had undergone total knee replacement and 99 had undergone total hip replacement for OA. Greater log-transformed concentrations of estradiol were associated with a lower incidence of knee replacement (hazard ratio [HR] 0.70, 95 confidence interval [95 CI] 0.50-0.96), and greater log-transformed concentrations of androstenedione were associated with a lower incidence of hip replacement (HR 0.70, 95 CI 0.52-0.93). In contrast, greater log-transformed concentrations of sex hormone binding globulin (SHBG) were associated with a higher incidence of hip replacement (HR 1.70, 95 CI 1.05-2.77). Conclusion A lower estradiol concentration is a risk factor for knee OA, while a lower androstenedione concentration and higher SHBG concentration are risk factors for hip OA in women. These findings suggest that circulating sex steroids have a role in the pathogenesis of OA, and that modifying these steroid concentrations may provide a potential strategy for the prevention and treatment of knee and hip OA. Copyright ? 2014 by the American College of Rheumatology.