We study the maximization of throughput and energy utilization in noisy wireless channels by adjusting packet length adaptively to network instant statistics. The optimal packet length adaptation (PLA) for throughput and energy utilization in wireless networks with and without re-transmission is respectively derived and developed. As more noises introducing more energy consumptions, the noises are equivalently regarded as lengthening of transmission distances. Therefore, an equivalent distance model of noisy channels is developed for more accurate estimation of the dissipated proportion in the residual energy such that further improvement of energy utilization and throughput is obtained. We integrate the PLA with the energy-proportional routing (EPR) algorithm for best balance of energy load. Therefore, performance metrics such as lifetime extension, throughput, and energy utilization are maximized even the distribution of channel noise is so un-predictable. Since the equivalent distance is dynamic, we believe that it is useful for network topology re-organization and will be useful in the future work of mobile ad-hoc networks.