Abstract. Autonomous scheduling deals with the problem - how to enable agents to schedule a set of interdependent tasks in such a way that whatever schedule they choose for their tasks, the individual schedules always can be merged into a global feasible schedule? Unlike the traditional approaches to distributed scheduling we do not enforce a fixed schedule to every participating agent. Instead we guarantee flexibility by offering a set of schedules to choose from in such a way that every agent can choose its own schedule independently from the others. We show that in case of agents with unbounded concurrency, optimal make-span can be guaranteed. Whenever the agents have bounded concurrency optimality cannot be guaranteed, but we present an approximation algorithm that ensures a constant make-span ratio.