The idea of distributed planning and problem solving as a reliable and effective methodology for air traffic control is introduced. Some basic concepts and terminology of distributed planning are first discussed, then a novel coordinator-coworker structure is described; this is a control mechanism that can dynamically reorganise the planning system in response to the task at hand. The information structure supporting such a control mechanism is made up of hierarchically organised distributed scratch pads. It also provides the basis for ‘self-repairing’ when individual nodes fail to operate, and allows for a graceful degradation of the whole system. The problems of air traffic control are analysed and the kernel design of the airborne processors in a location-centred co-operative planning system is described. The simulation-based planning process and its timing and selection considerations are explained. Three different organisational structures are implemented: the local centralised archi tecture, and the location-centred co-operative planning system with one- and two-level co ordinator-coworker hierarchies. The vehicle of the empirical studies, the distributed air traffic control testbed, is then introduced, and performance mea sures of the three organisational structures are compared in it. Finally, the design and the results of a series of experiments performed are discussed.