Recent advances in sensor, processor and airframe technologies allow coordination of large groups of autonomous unmanned aerial vehicles (UAV) today. Reconfiguration of the formation is sometimes necessary in order to accomplish a missionís objectives. A centralised solution to optimal reconfiguration may often be either impossible or intractable due to sensor, communication, physical, computational restrictions. Thus a distributed approach may be more appropriate to accommodate real-world scenarios. In this article we propose a novel distributed control method, which is divided into two modules: a leaderfollower module, which allows UAVs to keep a pre-specified formation, and a decision making module that allows UAVs to choose among various available formations in an optimum sense. UAVs choose the best formation to accomplish each part of the mission and retain this formation till the next way-point. The simulation presented uses a 5-leg mission and Parrot AR-drones are used as test-beds to demonstrate the usefulness of the proposed distributed controller.