Abstract: We consider the problem of multipath congestion control in the Internet. The aim is to take advantage of multiple paths diversity to achieve efficient bandwidth allocation and improve network efficiency. But there exist some potential difficulties when one directly uses the well-known network utility maximization model to design stable multipath congestion control algorithms for the alternative paths. In this paper, we propose a generalized multipath utility maximization model to consider the problem of joint routing and rate control, which can be reduced to specific models with different parameter settings. And then we develop a family of multipath dual congestion control algorithms which are stable in the absence of delays. We also derive decentralized and scalable sufficient conditions for a particular scheme when propagation delays exist in networks. The simulation results show that the proposed multipath dual congestion control algorithms with appropriate parameter settings can achieve stable resource shares while maintaining fairness among the involved users.