Optimization of industrial water networks made up of several water-using processes can reduce freshwater usage and wastewater production. A common feature of previous water network studies is that each water-using operation is assumed to have a single outlet stream. In this paper, a superstructure and a mathematical programing model are proposed for optimization of a water network with both single and two outflow water-using processes. A two outflow water-using process has two outlet streams with different impurity concentrations, in which one has a lower concentration and the other a higher concentration. Outlet stream with a high concentration can be reused in other water-using processes, regenerated in a regeneration system or discharged, while the outlet stream with a low concentration can only be reused in other water-using processes. These intricacies are accounted for in the mathematical model. For a two outflow water-using process, the main parameters are the allocation rates of flowrate and mass load which may be determined through consideration of the process's performance characteristics. The optimization methodology is demonstrated through a case study of a water network consisting of four single outflow water-using units, one two outflow water-using unit and a regeneration unit. (C) 2012 Elsevier B.V. All rights reserved.