Techniques for determining resources to use for peer-to-peer (P2P) communication are disclosed. In an aspect, a network entity may receive feedback information (e.g., resource usage information and/or channel state information) from P2P devices and may perform resource partitioning based on the feedback information to allocate some of the available resources for P2P communication. The allocated resources may observe little or no interference from devices engaged in wide area network (WAN) communication. In another aspect, P2P groups may perform resource negotiation via a WAN connection (e.g., with little or no involvement by the WAN) to assign the allocated resources to different P2P groups. In yet another aspect, a device may autonomously determine whether to communicate with another device directly or via a WAN, e.g., whether to initiate P2P communication with another device and whether to terminate P2P communication. In yet another aspect, a network entity may participate in resource negotiation by P2P devices.