Objective: To investigate the immunity markers related to nosocomial infection in children with sepsis. Methods: A retrospective study including 155 cases diagnosed as sepsis from September 2015 to June 2020 in children's intensive care unit (PICU) of Shanghai Children's Medical Center was conducted. According to the presence of nosocomial infection occurred in PICU, septic children were divided into two groups: no nosocomial infection and nosocomial infection group. The differences about helper T-cells 1 and 2 cytokines, T cells subgroup absolute count, the proportion of CD14+ human leukocyte antigen DR (CD14+HLA-DR), the proportion of regulatory T cells, pediatric risk of mortality Ⅲ (PRISM-Ⅲ), the treatment and outcome between the two groups were compared. Through propensity score matching (PSM), the disease severity and treatment of the two groups were matched to analyze the differences between the above indicators. Chi-square test or U test was used for comparison between groups. Receiver operating characteristic (ROC) curve was used to predict the occurrence of nosocomial infection. Results: There were 104 cases in no nosocomial infection group and 51 cases in nosocomial infection group. The first PICU-acquired infections occurred at (12±7) days after PICU admission. The most common PICU-acquired infections were pneumonia (26 cases, 51.0%) and bloodstream infections (15 cases, 29.4%). PRISM-Ⅲ of nosocomial infection group was significantly higher than that in no nosocomial infection group (8 (0-31) vs. 4 (0-17), Z=3 913.00, P<0.01).The proportion of using vasoactive drugs and invasive mechanical ventilation of nosocomial infection group was significantly higher (35.3% (18/51) vs. 10.6% (11/104), χ²=13.77, P<0.01; 86.3% (44/51) vs. 38.5% (40/104), χ²=31.51, P<0.01).The PICU length of stay of nosocomial infection group was significantly longer (20 (3-94) vs.7 (2-41) days, Z=4 585.50, P<0.01). The mortality of the nosocomial infection group was significantly higher than that of the group without nosocomial infection (29.4% (15/51) vs. 6.7% (7/104), χ²=14.45, P<0.01). Interleukin-6 and interleukin-10 of the nosocomial infection group were significantly higher than that in no nosocomial infection group (37.83 (2.23-7 209.99) vs. 13.45 (0.80~50 580.64) ng/L, Z=3 390.50, P=0.01; 10.42 (1.11-6 052.21) vs.4.10 (0.16-409.28) ng/L, Z=3 212.00, P=0.03). CD4+/CD8+ and the percentage of CD14+HLA-DR were significantly lower in the nosocomial infection group compared with the no nosocomial infection group (1.16 (0.44-4.96) vs. 1.61 (0.15-6.37), Z=1 955.00, P=0.01; 0.48 (0.08-0.99) vs. 0.67 (0.09-0.98), Z=1 915.50, P<0.01). After PSM, the percentage of CD14+HLA-DR of nosocomial infection group was significantly lower than that in no nosocomial infection group (0.44 (0.08-0.99) vs. 0.64 (0.09-0.98), Z=758.00, P=0.02). The ROC curve analysis of the percentage of CD14+HLA-DR in predicting nosocomial infection showed that the area under the curve was 0.642, the cut-off value was 0.39, and the 95%CI was 0.528-0.755. Conclusion: The level of the percentage of CD14+HLA-DR maybe is related to the occurrence of nosocomial infection in children with sepsis.