Diquat (DQ), an effective herbicide, has begun to threaten human health with the surge in the number of poisonings every year. Here, we present a powerful signature to improve the prognostic discrimination for patients with DQ poisoning. In this study, a total of 26 patients with acute DQ poisoning were enrolled. The median survival time in the death group was 43 hours. Undoubtedly, plasma concentration (PC), the blood cells (WBC), neutrophils (NE), aspartate aminotransferase (AST), alanine aminotransferase (ALT) and serum creatinine (Scr) in the death group were significantly higher and negatively correlated with survival time. And COX regression analysis suggested that PC, WBC, NE, ALT and Scr were the risk factors for poisoning death of DQ poisoning. Furthermore, risk factors were further developed as a robust signature (PC&Scr) to predict the prognosis of DQ poisoning. Survival analysis as well as time-dependent ROC analysis showed powerful predictive ability of prognostic signature in patients with DQ poisoning. A nomogram containing risk factors and signature could accurately predict the 2-day survival probability of individual DQ poisoning. The study shows that this signature with a powerful ability to predict DQ poisoning outcomes deserves popularizing in clinical practice.