Objective <p>The effects of gut microbiome and its metabolites on diabetic nephropathy (DN) have been inadequately elucidated. The aim of this study is to assess the causal effect of gut microbiome on DN and the mediated effect of metabolites by a two-step Mendelian randomization (MR).</p> Methods <p>Datasets of gut microbiome, metabolites, and DN were acquired in genome-wide association studies and screened for single nucleotide polymorphisms according to the underlying assumptions of MR. Subsequently, inverse variance weighted was used as the primary method for MR analysis to assess the causal effect of gut microbiome on DN and the mediated effect of metabolites. Finally, MR-Egger intercept, Cochran's Q test, and leave-one-out sensitivity analysis were used to assess the horizontal pleiotropy, heterogeneity, and robustness of the results, respectively.</p> Results <p>The MR analysis demonstrated that <i>Parabacteroides merdae</i> increased the genetic susceptibility to DN by reducing acetylcarnitine (C2) to propionylcarnitine (C3) ratio (mediated proportion 8.95%, mediated effect 0.024) and alpha-ketobutyrate to 3-methyl-2-oxovalerate ratio (mediated proportion 19.90%, mediated effect 0.053). MR Egger showed that these results lack horizontal pleiotropy (<i>p</i> ≥ 0.05). Cochran's Q and sensitivity analysis suggested these results had no heterogeneity (<i>p</i> ≥ 0.05) and were robust.</p> Conclusion <p>Our findings revealed the pathway by which <i>Parabacteroides merdae</i> increased the genetic susceptibility to DN by regulating acetylcarnitine (C2) to propionylcarnitine (C3) ratio and alpha-ketobutyrate to 3-methyl-2-oxovalerate ratio. It provides new genetic insights for understanding the pathogenesis of DN and related drug research.</p>