Long-term memory predictionextends the spatial displacement vector utilized in hybrid video coding by a variable time delay permitting the use of more than one reference frame for motion compensation. This extension provides improved rate-distortion performance. However, motion compensationin combinationwith transmissionerrorsleads to temporal error propagation that occurs when the reference frames at encoder and decoder Mer. In this paper, we present a framework that incorporatesan error estimate into rate-constrainedmotion estimationand modedecision. Experimentalresults with a Rayleighfading channelshow that long-term memory motion compensationsignificantly outperformssingle-frame prediction.