This paper proposes two energy-efficient techniques suitable for cooperative spectrum sensing for cognitive radios. Cooperative spectrum sensing collects information from different users to make final decisions. The simplest information to be collected is the decision of each user, and this method is analogous to voting. First of all, we propose a voting scheme based on users' own confidence. Users vote only when they are confident about their spectrum sensing results. Simulation results show that about 10% to 40% of transmission energy can be saved compared to traditional voting schemes. Secondly, we propose a cluster-collect-forward scheme to save energy spent on vote-collection and information exchange. This scheme relies on clustering for vote-collection, and combing and forwarding for information propagation. Our scheme shows significant energy saving from 65% to over 95% compared to the broadcasting scheme. Besides, the processing time is only about half of the broadcasting scheme. Finally, these two proposed techniques can be easily combined to further improve the overall performance.