Objective Few studies have examined the association between circulating sex steroid concentrations and risk of osteoarthritis (OA) in men with inconsistent results. Our aim was to examine whether concentrations of circulating sex steroid hormones were associated with the incidence of primary knee and hip arthroplasty for OA in a prospective cohort study. Design Two thousand four hundred and ninety four men from the Melbourne Collaborative Cohort Study (MCCS) had circulating sex steroid concentrations measured in blood samples drawn at recruitment (1990–1994) and stored in liquid nitrogen. The plasma concentrations of sex hormones, including dehydroepiandrosterone sulphate, androstenedione, testosterone, estradiol, androstanediol glucuronide, and sex hormone binding globulin, were measured. The incidence of total knee and hip arthroplasty for OA during 2001–2013 was determined by linking MCCS records to the Australian Orthopaedic Association National Joint Replacement Registry. Results One hundred and four men had knee and 80 had hip arthroplasty for OA over 10.7 (SD 3.8) years. Higher concentrations of androstenedione were associated with a decreased risk of total knee (hazard ratio (HR) 0.87, 95% confidence interval (CI) 0.77–0.98) and hip (HR 0.84 95% CI 0.71–1.00) arthroplasty for OA in overweight and obese men. No significant association was observed for the other measured hormones. Conclusion Low plasma androstenedione concentration is associated with an increased risk of both knee and hip arthroplasty for OA for overweight and obese men. While the findings need to be confirmed in other cohort studies, they suggest that circulating sex steroids may play a role in the pathogenesis of OA in men.