Objectives Observationally, testosterone is negatively associated with systemic inflammation, but this association is open to both residual confounding and reverse causality. Large-scale randomized controlled trials (RCTs), assessing exogenous effects, are presently unavailable. We examined the association of endogenous testosterone with well-established systemic inflammatory markers (white blood cell, granulocyte, lymphocyte and high-sensitivity C-reactive protein (hsCRP)) using a separate-sample Mendelian randomization analysis to minimize reverse causality. Methods A genetic prediction rule for serum testosterone was developed in 289 young Chinese men with mean age of 21.0, using selected testosterone-related SNPs (rs10046, rs1008805 and rs1256031). Multivariable linear regression was used to examine the association of genetically predicted serum testosterone with inflammatory markers among 4,212 older Chinese men from the Guangzhou Biobank Cohort Study. Results Genetically predicted testosterone was unrelated to white blood cell count (-0.01 109/L per nmol/L testosterone, 95% confidence interval (CI) -0.05 to 0.04), granulocyte count (-0.02 109/L, 95% CI -0.06 to 0.02), lymphocyte count (0.005 109/L, 95% CI -0.01 to 0.02) and hsCRP (-0.05 mg/L, 95% CI -0.15 to 0.06). Conclusion Our findings did not corroborate any anti-inflammatory effects of testosterone or corresponding potentially protective effects of testosterone on chronic diseases resulting from reduced low-grade systemic inflammation.