Recently, the number of the studies on the relationship between the total bilirubin level (TBL) and diabetic nephropathy (DN) is increasing, but their results were not consistent. Therefore, we performed a meta-analysis to analyze the relationship between TBL and the risk of DN.We searched 5 databases before October 31, 2016, and reviewed the reference list of relevant articles. The fixed or random-effects model was used to pool risk estimates. We conducted the dose-response meta-analysis to evaluate the relationship between TBL and the risk of DN.Our meta-analysis showed that TBL in the DN group was lower than that in diabetes without the kidney disease (NDN) group (standard mean difference [SMD]: -0.63, 95% CI: -0.80, -0.46). The result of each subgroup also showed that TBL in the DN group was lower than that in the NDN group. The result of meta-regression indicated that duration of diabetes mellitus might be the source of heterogeneity. Our meta-analysis also showed that there was a significant negative relationship between TBL and the risk of DN (OR: 0.86, 95%CI: 0.82, 0.90). The results of subgroup analysis were similar to those of SMD; no sources of heterogeneity were detected by meta-regression. Sensitivity analysis indicated that the results were robust. We observed a linear association between TBL and the risk of DN, and there was a negative dose-response association between TBL and the risk of DN.In conclusion, bilirubin may be used as a biomarker of DN. It helps early diagnosis and effective therapeutic strategies on DN.