To analyze the risk factors of patients with trauma in intensive care unit (ICU), a new warning scoring system is established for predicting the incidence of sepsis in traumatic patients; and to provide a new simple method of clinical score, which could provide a reference for clinical prevention and treatment of sepsis. The clinical data of 591 patients with trauma in the ICU of the Army Specialized Medical Center of Army Medical University and Affiliated Hospital of Zunyi Medical University from January 2012 to December 2017 were retrospectively analyzed. The patients were divided into sepsis group (n = 382) and non-sepsis group (n = 209) according to their clinical outcome. The basic clinical data of all ICU trauma patients were collected, and the differences in gender, age, underlying diseases, and vital signs, critical illness scores, blood culture results and laboratory biochemical examinations within 24 hours of ICU admission between the two groups were analyzed. Univariate Logistic regression analysis was used to screen the related factors leading to sepsis. The indexes with P < 0.12 analyzed by univariate Logistic regression analysis were included in multivariate Logistic regression analysis. The risk factors of sepsis in traumatic patients were screened and assigned, and the total score was sepsis early warning score. The receiver operating characteristic (ROC) curve was plotted to evaluate the predictive value of the warning score of sepsis in patients with trauma. The incidence of sepsis in ICU trauma patients was 64.6% (382/591), and the ICU mortality was 10.5% (40/382). The traffic accident was a common cause of ICU trauma patients. Compared with non-sepsis patients, Glasgow coma score (GCS), proportion of past history, red blood cell (RBC), platelet (PLT), albumin (Alb) were lower in patients with sepsis, and body temperature, pulse, acute physiology and chronic health evaluation II (APACHE II), sequential organ failure assessment (SOFA), injury severity score (ISS), new injury severity score (NISS), fraction of inspired oxygen (FiO2), blood sodium, activated partial thromboplastin time (APTT), prothrombin time (PT), procalcitonin (PCT), C-reactive protein (CRP) levels were higher, blood transfusion, central venous catheterization, mechanical ventilation, shock, multiple organ dysfunction syndrome (MODS), open injury and multiple injuries were more common, the duration of mechanical ventilation, ICU days and total hospital days were longer, and all the differences were statistically significant. Most of the traumatic patients with sepsis were undergone with multiple trauma. Compared with non-sepsis patients, the proportion of multiple position trauma was significantly higher than patients without sepsis. And most traumatic patients were insulted in head, face and neck. The risk factors were screened by univariate and multivariate Logistic stepwise regression analysis, the indexes into the regression model were pulse > 100 bpm [odds ratio (OR) = 1.617, 95% confidence interval (95%CI) = 0.992-2.635, P = 0.044], APTT > 36 s (OR = 2.164, 95%CI = 1.056-4.435, P = 0.035), shock (OR = 1.798, 95%CI = 1.056-3.059, P = 0.031), mechanical ventilation (OR = 5.144, 95%CI = 2.302-11.498, P < 0.001), APACHE II > 21 (OR = 3.348, 95%CI = 1.724-6.502, P < 0.001), NISS > 25 (OR = 3.332, 95%CI = 1.154-9.624, P = 0.026), assigning scores were 0.5, 1.0, 0.5, 1.5, 1.5, 1.5, respectively, which were included in the new warning score of sepsis. ROC curve analysis showed that the area under ROC curve (AUC) of warning score for predicting sepsis in patients with trauma was 0.782, which was significantly higher than the APACHE II (AUC = 0.672), APTT (AUC = 0.574) and NISS (AUC = 0.515) with significant difference (all P < 0.01). When the cut-off value of sepsis warning score was 4.0, the sensitivity and specificity were 71.7% and 61.9%, respectively. Close monitoring and stabilization of vital signs of traumatic patients within 24 hours of ICU admission and reduction of unreasonable invasive mechanical ventilation time are expected to reduce the incidence of sepsis in traumatic patients. New warning score of sepsis consisted of six factors: pulse, APTT, shock, mechanical ventilation, APACHE II and NISS. Rational use of warning score of sepsis would help us to assess the prognosis of traumatic patients more easily and effectively, and the predicted effect is much better than APACHE II, APTT and NISS.