Analysis of wireless systems commonly assumes single-user detection at the receivers. Interference is typically treated as noise. On the other hand, multiuser detection has long been taking advantage of interference cancellation (IC) to increase capacity. We exploit IC by optimal rate selection. Transmission rates are collaboratively optimized to maximize the benefit of IC. A link reduces its rate, if that enables IC to significantly boost the SINR on other links. We provide a complexity analysis and an integer programming model to find the optimal IC pattern. Simulation results indicate that throughput improvement is over 30% in low SINR regimes.