||With the penetration of broadband Internet and Wi-Fi-capable electronic products to homes, there has been a growing popularity to set up WLAN in homes, the so-called home WLAN. As opposed to traditional WLAN networks (such as a campus or company), these home WLANs are usually set up by network non-specialists, uncoordinated, and of unplanned topology. Therefore, the access points (APs) from different homes may interfere with each other, leading to unsatisfactory throughput performance. We study in this paper distributed channel assignment in an uncoordinated home network, so that APs can self-configure its operating channel to minimize interference with each other. We first discuss the weaknesses of a traditional channel assignment scheme, Least Congested Channel Search (LCCS), and formulate an assignment problem to overcome them. The problem is NP-hard, and hence we propose a scalable algorithm termed Client-Assisted Channel Assignment Optimization (CACAO) for home WLAN. In CACAO, an AP makes use of the feedback of the client’s local traffic information to make channel assignment decision. This leads to a better knowledge on the network condition, and a channel with the least interference. In CACAO, an AP continuously improves its channel selection until some optimum is achieved. CACAO is efficient, fully distributed, and simple to implement. Besides simulation, we have also prototyped CACAO and conducted proof-of-concept experiments. Both simulations and experimental measurements confirm that CACAO effectively reduces interference among different home WLANs, achieves fast convergence time, and substantially improves user throughput (doubly the throughput in some cases).