Cooperative spectrum sensing is an advanced technology to improve the utilization of scarce radio spectrum resources. However, there may exist some malicious secondary users (SUs) who are unwilling to cooperate, i.e. provide false sensing reports. Such malicious behavior, if unchecked, the sensing performance of cooperative spectrum sensing will be severely degraded. In this paper, we propose a trust weighted cooperative spectrum sensing scheme to identify malicious SUs and mitigate their harmful effect on sensing performance. This scheme provides a reliable approach to evaluate trust values for SUs. Depending on the trust values, this scheme can detect malicious SUs and filter out their false spectrum sensing results in the process of sensing data fusion. Finally, simulation results validate the effectiveness of the proposed scheme.