Objective: Cumulative evidence from observational studies showed an inverse association between smoking and osteoarthritis (OA), but this association could be confounded by obesity. This study aimed to decipher the causal effect of smoking on osteoarthritis risk from a genetically informed perspective. Methods: We performed a two-sample univariable and multivariable MR to evaluate the independent effect of smoking on OA risk. Summary-level data were obtained from the largest available genome-wide association studies (GWASs) of smoking initiation, body mass index (BMI) and OA. Genetic variants predicted the exposure were selected as instruments from the respective GWAS. Results: Genetically liability for smoking initiation had an effect estimate consistent with increased risk for overall OA (odds ratio (OR)=1.61, 95% confidence interval (CI)=1.43–1.81, P=7.50 × 10−15), hip OA (OR=1.62, 95%CI=1.29–2.02, P=2.93 × 10−5), knee OA (OR=1.54, 95%CI=1.29–1.84, P =1.80 × 10−6) and hip and/or knee OA (OR=1.56, 95%CI=1.34–1.80, P=3.63 × 10−9), respectively. We also found that genetic liability of BMI was significantly associated with OA risk and the OR per genetically predicted 1 kg/m2 increase of BMI ranged from 2.19 to 2.64. Additionally, multivariate MR analysis revealed a strong evidence for an effect of smoking initiation on the risk of overall OA (OR=1.45, 95%CI=1.31–1.61, P=3.69 × 10−12) and its subtypes after controlling for BMI. Conclusion: Our findings support an independent deleterious causal effect for smoking upon OA risk, which further strengthen the importance of smoking cessation interventions and obesity management in the general population, in order to lessen the huge burden of OA in the global aging era.