Abstract: OBJECTIVE To investigate the direct economic loss due to nosocomial infection by using propensity score matching（PSM） and generalized linear model（GLM） so as provide guidance for development of control measures. METHODS A retrospective survey was conducted for the hospitalized information of all of the patients who were discharged from Shenzhen Hospital, Chinese Academy of Sciences from Jun 2016 to Nov 2018, the patients were divided into the nosocomial infection group and the non-nosocomial infection group, and the length of hospital stay and increase of hospitalization cost of the case group and the control group were studied by means of PSM and GLM. RESULTS After PSM, the median length of hospital stay of the nosocomial infection group was increased by 14.00 days, 2.00 times the length of hospital stay of the non-nosocomial infection group（Z=-21.485,P＜0.001）.The median hospitalization cost of the nosocomial infection group was increased by 17 264.00 yuan, 1.57 times the hospitalization cost of the non-nosocomial infection group（Z=-13.576,P＜0.001）, the result of PSM was consistent with that of the GLM.The extended length of hospital stay of the hospitalized patients with infection of neurosurgery department was the longest, reaching to 31.00 days（Z=-8.225,P＜0.001）.The increased hospitalization cost of the ICU patients with infection was the most, reaching to 80096.00 yuan（Z=-6.371,P＜0.001）. CONCLUSION The nosocomial infection may result in remarkable extension of length of hospital stay and increase of hospitalization cost, and it is necessary to focus on the control of nosocomial infection in the neurosurgery department and ICU and develop targeted measures.