Abstract: Water demand simulation is an important basis of regional water resources planning and management. Technologies of Choleskey factoring, Box-Cox data transferring and Monte-Carlo simulation are combined to establish a multi-dimension stochastic simulation model of regional water demand. And this model can realize long time-series simulation and express the correlativity between monthly water demands. The middle-lower reaches of HanJiang River was taken as study object. Results showed that the relative errors of mean value, variance and correlation coefficient of the simulated water demand series had least errors. This study presented a new method of water demand simulation with application value.