A cooperative task allocation and search algorithm is proposed to nd and localize a group of ground-based moving targets using a group of Unmanned Air Vehicles (UAVs) working in a decentralized manner. It is assumed that targets have RF emissions. By using an algorithm including Global Search (GS), Approach Target (AT), Locate Target (LT), and Target Reacquisition (TR) modes, UAVs cooperatively search the entire parts of a desired area, approach to the detected targets, locate the targets, and search again to nd the targets that stop transmitting their RF emissions during the localization process, respectively. In the GS mode, UAVs utilize a cost function to select the best zone for search. In the LT mode, each UAV performs a circular motion around the target and uses an extended Kalman lter to estimate the target position. Furthermore, a fuel tanker is considered to provide fuel for UAVs during the ight. Therefore, two more operating modes including Approach to Fuel Tanker (AFT) and fueling (FUE) are added to the operating modes. Before switching to the AFT mode, UAVs take turn using a fueling decision function. In the AFT mode, the future position of the fuel tanker is predicted by UAVs to reduce the approach time.