The key question need to be answered about relay selection in a large-scale randomly distributed wireless network is that how to select a relay energy efficiently under a throughput (i.e., effective data rate) constraint with a reasonable complexity. It is not practical to collect instantaneous channel state information of all relays for relay selection because of excessive control packet transmission and useless energy consumption because to hold a relay in active mode to participate relay selection process needs significantly more power consumption than to hold in sleep mode. In this paper we propose a new energy-efficient relay selection technique for large-scale randomly distributed wireless networks for which an active relay region is introduced where only the relays in the region are taken into consideration for relay selection. Then, we propose a simple relay selection technique, which is only based on the channel state information of source-relay channels of the relay subset to reduce the complexity and signaling overhead. The proposed relay selection technique increases energy efficiency at an acceptable complexity level without jeopardizing the effective data rate. We investigate the optimum size of the active relay region for energy efficient in which relays perform decode-and-forward technique. A detailed and realistic power consumption model is studied by taking all possible sources of power consumption into consideration, i.e., data transmission, signaling packets transmission, and circuit power consumption during all phases of the communication. We show that the proposed technique is more energy efficient and provides better effective data rate in comparison to the reference techniques.