Preemptive. A single queue of ready processes is maintained, and a fixed time quantum is assigned to each running process. After the quantum expires, the process is blocked and put at the end of the queue, and the next in the line is dispatched. This method is obviously very fair, and provides good responsiveness to short processes. The drawback is that the length of the time quantum must be carefully tuned, otherwise throughput will suffer. If it's too short, there may be excessive process switching overhead, if too long we are back to the lack of fairness problem of FCFS. This technique prevents starvation.