Controlling the preemption behavior in real-time systems can have beneficial impacts in multiple contexts as it can decrease the processor utilization, reduce the energy consumption or even enable the schedulability of the system. In this paper we study the preemption behavior of sporadic task systems scheduled using the Fixed Priority Scheduling (FPS) policy, and evaluate the feasibility of preemption control using CPU frequency scaling. We show that offline preemption control using CPU frequency scaling is difficult for sporadic task systems, and we propose an online heuristic algorithm, of linear complexity, to control the number of preemptions in a sporadic task system. Evaluation results show that online CPU frequency scaling is an attractive approach for preemption control in sporadic task systems.