In a cognitive radio (CR) network, the system throughput and the probability of collision between primary users (PUs) and CR users are directly influenced by the frequency of spectrum sensing. This paper focuses on adaptively scheduling spectrum sensing such that negative impacts to the performance of the CR network are minimized. Based on an in-depth analysis into the spectral usage patterns of PUs, an adaptive spectrum detecting algorithm is proposed. By introducing a "control factor", the proposed algorithm can adaptively schedule the spectrum sensing, while maintaining the CR system stability, so as to increase spectral utilization efficiency, as well as reducing the probability of collision between PUs and CR users. Therefore, the energy consumption of the CR system is reduced. Simulations show that the proposed algorithm can effectively increase the system throughput, with minimal interference to primary users while ensuring the system stability. Meanwhile, it is shown that the proposed algorithm has low implementation complexity for practical applications.