In a WiFi deployment with multiple access points, optimizing the way each client selects an AP from amongst the available choices, has a significant impact on the realized performance. When two or more such multi-AP networks are deployed in the same region, APs from different networks can cause severe interference to one another. In this paper, we study how inter-network interference effects the intra-network association optimization and propose a cooperative optimization scheme to mitigate the interference. We model the interference between multiple overlapping WiFi deployments, determine the information that networks need to share, and formulate a non-linear program that each network can solve for optimal proportional-fair association of clients to APs. Assuming a sum of log rates utility function, we apply a known 2+∈ approximation algorithm for solving the NP-hard problem in polynomial time. We evaluate the performance gain through large-scale simulations with multiple overlapping networks, each consisting of 15-35 access points and 50-250 clients in a 0.5×0.5 sq.km. urban setting. Results show an average of 150% improvement in random deployments and upto 7× improvements in clustered deployments for the least-performing client throughputs with modest reductions in the mean client throughputs.