【Objective】To analyze the risk factors of progression to end-stage renal disease（ESRD）in patients with diabetic kidney disease（DKD），and screen the high-risk population for early prevention.【Methods】The clinical data of 231 patients with diabetic nephropathy in our hospital were collected and followed up for 3 years. According to whether ESRD occurred，they were divided into non-progressing ESRD group（133 cases）and ESRD group（98 cases）. Classification tree model was used to analyze the risk factors related to ESRD，and the high-risk population was screened by node gain analysis.【Results】Four important explanatory variables were screened out by the classification tree model from the candidate variables related to early renal damage，including apolipoprotein B（ApoB），gender，diabetic retinopathy，systemic blood pressure（SBP）. ApoB level was an important factor for DKD progression. For DKD patients with the chronic kidney disease （CKD）3~4 stages with ApoB > 1.14 mmol/L，the probability of progression to ESRD for 3 years was 75.0 %，and if at the same time with diabetic retinopathy，the probability was 79.7 %.【Conclusion】The classification tree model can analyze the risk factors of progression to ESRD in DKD patients effectively，to identify the characteristics of high-risk populations.