Diabetic nephropathy remains a common cause of end-stage renal failure and its associated mortality around the world. Sphingosine 1-phosphate (S1P) is a multifunctional lipid mediator and binds to HDL via apolipoprotein M (ApoM). Since HDL has been reported to be epidemiologically associated with kidney disease, we attempted to investigate the involvement of the ApoM/S1P axis in the pathogenesis/progression of diabetic nephropathy. In type 2 diabetic patients, the serum ApoM levels were inversely correlated with the clinical stage of diabetic nephropathy. The decline in the eGFR over a 5-year observation period proceeded more rapidly in subjects with lower serum ApoM levels. In a mouse model of streptozotocin-induced diabetes, deletion of ApoM deteriorated the phenotypes of diabetic nephropathy: the urinary albumin and plasma creatinine levels increased, the kidneys enlarged, and renal fibrosis and thickening of the basement membrane progressed. On the other hand, overexpression of ApoM ameliorated these phenotypes. These protective effects of ApoM were partially inhibited by treatment with VPC23019, an antagonist of S1P1 and S1P3, but not by treatment with JTE013, an antagonist of S1P2. ApoM/S1P axis attenuated activation of the Smad3 pathway, while augmented eNOS phosphorylation through the S1P1 pathway. Moreover, ApoM/S1P increased the SIRT1 protein levels and enhanced mitochondrial functions by increasing the S1P content of the cell membrane, which might cause selective activation of S1P1. ApoM might be a useful biomarker for predicting the progression of diabetic nephropathy, and the ApoM/S1P-S1P1 axis might serve as a novel therapeutic target for preventing the development/progression of diabetic nephropathy.