Wireless media access protocols for a single channel can typically be categorized as either token-based or multiple access. For reasons we explain in the next section, we choose the multiple access approach. 1 Our work is based on MACA, a Multiple Access, Collision Avoidance protocol first proposed by Karn [9] and later refined by Biba [3]. Using packet-level simulations of the wireless network to guide our design, we suggest several modifications to MACA. We call the resulting algorithm MACAW, in recognition of its genealogical roots in Karn’s original proposal Our design is based on four key observations. First we observe, following Karn [9] and others [4, 12], that the relevant contention is at the receiver, not the sender. This renders the carrier sense approach inappropriate. Second, we note that, in contrast to Ethernets, congestion is location dependent; in fact, the first observation is irrelevant wit bout the second. Third, we conclude that, to allocate media access fairly, learning about congestion levels must be a collective enterprise. That is, the media access protocol should propagate congestion information explicitly rather than having each device learn about congestion independently. Fourth, the media access protocol should propagate synchronize ation information about contention periods, so that all devices can cent end effectively. In particular, this means that cent ention for bandwidth should not just be initiated by the sending device. While our proposed protocol provides enhanced performance (as compared to MACA), we hasten to note that it is merely an initial attempt to deal with these challenges; there are many remaining unresolved design issues.