Water supply networks are generally modelled for a day or longer time period. Under most circumstances, it is normal to omit the dispersive movement of chlorine and to assume that the convective transport dominates over dispersion. However, during night time operations the velocities are significantly decreased. This radical change in the velocity may therefore lead to the following question: can the available convection based quality models predict the chlorine concentrations accurately in time and space? An improved water quality model implies accurate estimation of concentrations at nodes in time and space at all levels of branching. To enhance the utility of water quality models dispersion may also be taken into account. In this paper, an effort that Led to the development of an unsteady 2-D convective-dispersive model is presented, and the model output is compared with the experimental results. (C) 1999 Elsevier Science Ltd. All rights reserved.