For many water-intensive processes, water reuse can reduce water consumption as well as effluent generation. Process integration approach based on graphical pinch methodology for targeting and water network synthesis is often employed. The integrity of water network design to achieve the minimum water targets is highly sensitive to the availability of reliable process data. Existing network design process, however, assume that process data are fixed and well-defined, whereas the actual operating conditions such as water flowrate and the corresponding mass loads may fluctuate over time. These fluctuations in processing conditions can lead to process disruptions and product quality problems. This work demonstrates the use of Monte Carlo simulation in assessing the vulnerability of water networks to noisy mass loads. A case study illustrates the procedure of selecting the most robust network configuration from three alternative designs that achieve comparable water savings. (c) 2006 Elsevier Ltd. All rights reserved.