Orthogonal frequency division multiplexing (OFDM) has been investigated as an enabling technology for future wireless communications such as ad hoc, mesh and sensor networks. However, prior works on bit-loading lack consideration of the constraints on energy and computing facility in sensor networks. In this paper, we suggest an adaptive bit allocation algorithm for a frequency selective fading channel environment which exploits channel state information obtained through a feedback channel. The proposed algorithm significantly reduces computational complexity and satisfies the power budget. Also, its throughput is comparable to the optimum solution. Simulation results support the claim stated.