In this paper, using the Dempster–Shafer theory (DST) of evidence, a new decision criterion is proposed which can quickly classify airborne objects without any a priori knowledge, whose data are laced with environmental noise characteristics, within 10seconds (10s) from the time it is detected. Kinematic parameters of an airborne object received from radars are used to classify it into one of the six classes, which include three levels of ballistic target discrimination, aerodynamic, satellite and unknown. The DST is chosen as it can suitably handle the element of uncertainty, limited a priori data and short observation times that exist with the data acquired for the purpose of classification. The focus of the work is on ballistic targets in a theater of war. The approach is compared with the popularly known k-NN and decision tree techniques and is found to perform better with the chosen data sets. This approach is tested using both real flight test data and simulated data.