In the cognitive radio networks (CRNs), efficient spectrum sensing is a key enabling techniques for the effective spectrum exploration. However, Majority of the existing works on spectrum sensing is conducted the spectrum sensing serially in cognitive radio networks, which may bring about the sensing delay problem. In this paper, in order to make full use of the available channel resources and improve the spectrum exploitation, we propose a channel sensing order optimization for secondary users in the scenario that the CRN has a wide spectrum band with multiple channels. Firstly, we extend the preliminary screening mechanism to select the channel set for the secondary users to detect the spectrum hole with descending the product of the idle probability of the channels and channel throughput. Furthermore, the spectrum sensing order is decided according to the Markov prediction technology. Numerical results show that the proposed algorithm can improve the spectrum sensing efficiency and increase the system throughput.