Header compression techniques theoretically can improve the speech quality of multimedia applications. However it introduces a new impairment directly related to the loss of packets, that can impact negatively on the quality. Its deployment over wireless mesh networks needs special treatment due to the usually high loss rates of such a network. This work shows an evaluation of an alternative approach for header compression over wireless mesh networks. Experiments by means of stochastic simulation showed that under some network characteristics the proposed approach shows significant improvements in the packet loss.