Source-sink problems occur in various process systems engineering applications, such as the design of water reuse systems and resource conservation networks. Such problems may be solved using alternative approaches, including graphical pinch analysis and linear programming. In most cases, it is normally assumed that the stream flow rates and quality levels are deterministic. This paper presents a fuzzy mathematical programming model for the synthesis of water networks when the model parameters exhibit fuzzy uncertainties. The modeling approach is illustrated for three water network case studies adapted from literature. These illustrative examples cover concentration- and property-based stream quality indices, as well as single- and multiple-plant water integration.