Objective To explore the risk factors of multiple organ dysfunction syndrome (MODS) in severe trauma patients, put forward a new warning scoring system of MODS, and to provide a more accurate scoring method for doctors to judge the clinical condition and prognosis of patients. Methods Clinical data of 342 patients with severe trauma admitted to intensive care unit (ICU) of the Affiliated Hospital of Zunyi Medical College and Daping Hospital of the Third Military Medical University from January 1st, 2015 to December 31st, 2016 were retrospectively analyzed. The patients were divided into MODS groups (n = 251) and non-MODS group (n = 91) according to clinical outcomes. The clinical data of patients, including gender, age, heart rate (HR) and blood pressure within 24 hours after admission to the hospital, indicators of blood routine and blood biochemistry, severity of disease, severity of trauma, whether received the emergency intubation or surgery within 24 hours or not, whether developed sepsis or acute respiratory distress syndrome (ARDS) during hospitalization, were recorded, and univariate analysis was conducted. The indicators with statistical significance found by univariate analysis were enrolled in multivariate Logistic regression analysis, and the risk factors for MODS in patients with severe trauma were screened and assigned, and the final total score was MODS warning score. Receiver operating characteristic (ROC) curve was plotted to evaluate MODS warning score for predicting the occurrence of MODS in patients with severe trauma. Results Compared with non-MODS group, HR, Na+, serum creatinine (SCr), activated partial thromboplastin time (APTT), injury severity score (ISS), new injury severity score (NISS), acute physiology and chronic health evaluation Ⅱ (APACHEⅡ) score and sequential organ failure assessment (SOFA) score in MODS group were significantly increased, pH value, red blood cell (RBC), platelet (PLT), albumin (Alb) and Glasgow coma score (GCS) were remarkably decreased, and multiple injury, the patients with shock at admission, blood transfusion, central venous catheter, emergency intubation and infection were also increased, and more patients suffered from sepsis and ARDS. Multivariate Logistic regression analysis showed that the number of injured places equal or more than 2, shock at admission, APACHEⅡ score ≥ 15, SOFA score ≥ 4 and APTT > 40 s were risk factors for MODS in patients with severe trauma, with total MODS warning score of 7.5. ROC curve analysis showed that the area under ROC curve (AUC) of MODS warning score for predicting MODS in patients with severe trauma was 0.822, which was significantly higher than that of APACHEⅡ score (AUC = 0.698, P < 0.001), APTT (AUC = 0.693, P < 0.001) and SOFA score (AUC = 0.770, P = 0.025). When the cut-off value of MODS warning score was 2.5, the sensitivity was 61.35%, the specificity was 90.11%, and Youden index was 0.515. Conclusions MODS warning score is composed of five factors, including the number of injured places, shock at admission, APACHEⅡ score, SOFA score and APTT, which could be regarded as early warning score system for predicting MODS in patients with severe trauma. MODS warning score can be more comprehensive and timely to assess the possibility of MODS and prognosis of patients with severe trauma, and the prediction result is better than the single use of APTT, APACHEⅡ or SOFA score.