BACKGROUND: The reported association of smoking with risk of undergoing a total joint replacement (TJR) due to osteoarthritis (OA) is not consistent. We evaluated the independent association between smoking and primary TJR in a large cohort. METHODS: The electronic records of 54,288 men and women, who were initially recruited for the Second Australian National Blood Pressure study, were linked to the Australian Orthopaedic Association National Joint Replacement Registry to detect total hip replacement (THR) or total knee replacement (TKR) due to osteoarthritis. Competing risk regressions that accounted for the competing risk of death estimated the subhazard ratios for TJR. One-way and probabilistic sensitivity analyses were undertaken to represent uncertainty in the classification of smoking exposure and socioeconomic disadvantage scores. RESULTS: An independent inverse association was found between smoking and risk of THR and TKR observed in both men and women. Compared to non-smokers, male and female smokers were respectively 40% and 30% less likely to undergo a TJR. This significant association persisted after controlling for age, co-morbidities, body mass index (BMI), physical exercise, and socioeconomic disadvantage. The overweight and obese were significantly more likely to undergo TJR compared to those with normal weight. A dose-response relationship between BMI and TJR was observed (P < 0.001). Socioeconomic status was not independently associated with risk of either THR or TKR. CONCLUSION: The strengths of the inverse association between smoking and TJR, the temporal relationship of the association, together with the consistency in the findings warrant further investigation about the role of smoking in the pathogenesis of osteoarthritis causing TJR.