Introduction: Dengue fever (DF) is common in Asia. Dengue hemorrhagic fever (DHF) occurs predominantly in children less than 16 years old. In practice, the results of diagnostic tests, whenever available, are rarely received in time to be useful for immediate treatment in endemic regions. Our goal is to identify simple laboratory features to discriminate DF from non-dengue fever (NDF).Methods: There are two parts used to investigate the simple way discriminating DF. First, multivariate discrimination analyses were applied to those three statistical techniques of rotate axis, logistic regression and principle component analysis in detecting their effects. Second, three scaling schemes of dichotomy with coding 2 and 1, 1 and 0 as well as of polytomy coding 2, 1 and 0 on WBC and PLT by adding category scores were applied to verify the best model to predict Dengue fever. In this model, lower values tend to be DF.Results: WBC and PLT are the discriminatory variables that enable distinction between DF and NDF. Using the area under the ROC curve (AUC) to verify the efficacies of the three statistical prediction methods and coding schemes, we recommend that polytomy with a cut-point <= 2 by the coding scheme from 2 to 0 can be applied to clinical practice because it is easier and faster than other available prediction methods.Conclusions: We suggest using a polytomous coding scheme with a cut-point <= 2 as a technique for rapidly detecting DF. Further studies should be performed in different countries and areas in the future.