Abstract: OBJECTIVE To study the characteristics of neonatal nosocomial infection, the classification tree model was used to analyze the main risk factors of neonatal nosocomial infection, identify the susceptible groups, and provide references for corresponding clinical intervention measures. METHODS Personal information and clinical data of 5303 neonates admitted from Jun. 2016 to Nov. 2018 were retrospectively analyzed, including gender, birth weight, length of hospital stay, device utilization, parenteral nutrition and antibiotic use, to understand the status of infection, and the classification tree model was used to analyze the risk factors of nosocomial infection. RESULTS During the observation period, 92 patients had 98 cases of nosocomial infection in the 5303 neonates, the infection rate and case rate were 1.73% and 1.85%, respectively. The blood system (63.27%) and gram-positive bacteria(59.46%) were the most common infection site and pathogens. Univariate analysis and classification tree model analysis showed that the hospitalization time, birth weight and device utilization were the independent risk factors of neonatal nosocomial infection. The infection rate of neonates with length of hospital stay >14 days, birth weight lower than 1500g and device utilization was 27.48%, and they were the most susceptible individuals. CONCLUSION Neonates with low birth weight, device utilization and long hospitalization time are at high risk of nosocomial infection in neonatal wards. They should be regarded as the focus of attention and protection. Strengthening the management of this population has great significance in reducing nosocomial infection in neonates.