Introduction: Observational studies have found lower concentrations of plasma bilirubin in current smokers compared with former and never smokers. However, whether there is a causal relationship between smoking and bilirubin is unknown. In a Mendelian randomization analysis, we tested the hypothesis that higher tobacco consumption is causally associated with lower concentrations of plasma bilirubin. Methods: We genotyped 103 557 individuals aged 20-100 years from the Copenhagen General Population Study for the CHRNA3 rs1051730 genotype, known to be associated with higher tobacco consumption. Tobacco consumption was defined as daily and cumulative tobacco consumption. Results: In observational multivariable-adjusted analyses, a 10 g/day higher daily tobacco consumption was associated with a 0.28 μmol/L (95% confidence interval = 0.20 to 0.35) lower concentration of plasma bilirubin in current smokers, and a 10 pack-year higher cumulative tobacco consumption was associated with a 0.19 μmol/L (0.17 to 0.21) lower concentration of plasma bilirubin in former and current smokers. Using the CHRNA3 rs1051730 genotype as a proxy for daily and cumulative tobacco consumption, the difference in plasma bilirubin per T-allele was -0.12 μmol/L (-0.23 to -0.002) in current smokers and -0.09 μmol/L (-0.15 to -0.01) in current and former smokers combined. Furthermore, observationally bilirubin concentrations increased with time from smoking cessation in former smokers. Conclusion: Higher daily and cumulative tobacco consumption were associated with lower concentrations of plasma bilirubin in observational and genetic analyses, suggesting that the association is causal. Implications: Our results are compatible with two possible interpretations of previous observational studies, either that bilirubin is a mediator of smoking-induced respiratory disease or that the association between plasma bilirubin and respiratory disease stems from residual confounding because of smoking. Future studies should examine whether bilirubin is a causal risk factor for respiratory disease, or merely a marker of smoking status.