ZigBee in conjunction with IEEE 802.15.4 is considered the de facto standard for WSNs. The routing protocol for ZigBee mesh networks is AODV, an on-demand routing protocol which has been extensively used. However, it presents critical problems when it is used in large dense networks, mainly related to the high number of messages necessary to perform routing tasks, which causes a very high overhead. The high number of packet collisions, data loss, and its implicit energy expenditure, makes AODV unsuitable for certain applications that require efficient, fast and reliable communications. This study proposes the use of a novel fuzzy-logic-based metric to be used in the decision making process of AODV. This metric evaluates important node features during route creation and selects, hop by hop, the best option in order to obtain energy and delay efficient routes. The experiments with the improved AODV, AODV-FL show a reduction in the communication delay, number of packets and overhead, improving route efficiency and reducing packet overload. © 2011 IEEE.