Objective To evaluate the feasibility of developing hospital acquired pneumonia (HAP) risk warning model in critically ill patients based on genomic copy number polymorphisms (CNPs) of the genes encoding human neutrophil peptides 1-3 (DEFA1/DEFA3).Methods Seventy-seven HAP patients (group HAP) and 109 non-HAP patients of matched age and sex in intensive care unit (ICU) (group NHAP) were enrolled in the study.The genomic CNPs of DEFA1/DEFA3 was determined by realtime quantitative polymerase chain reaction after extracting DNA from peripheral blood samples.The source of patients,condition of endotracheal intubation within 24 h after admission to ICU,Acute Physiology Score,Acute Physiology and Chronic Health Evaluation Ⅱ score,Sequential Organ Failure Assessment score,mechanical ventilation time,length of hospital and ICU stay and outcomes were obtained.The predictive model was developed using logistic regression through combining DEFA1/DEFA3 copy numbers and clinical characteristics (Acute Physiology Score and source of emergency) within 24 h after admission to ICU.The receiver operating characteristic curve was used to evaluate the predictive efficacy of the model.Results The copy numbers of DEFA1/DEFA3 were significantly lower in HAP group than in NHAP group (P ＜0.05).The area under the receiver operating characteristic curve of the predictive model developed through combining the DEFA1/DEFA3 copy numbers with clinical characteristics was 0.789 (95％ CI 0.724-0.854) when the model was used for predicting HAP.Conclusion CNPs of DEFA1/DEFA3 can be used to develop the HAP risk warning model in critically ill patients.