Water supply system model is an important method to ensure the water quality safety in water supply system, however, the water quality value calculated by water quality model affected by the uncertainty of input parameters is uncertain. In order to study the influence of the basic water consumption amount per node, water loss coefficient, main body water response coefficient and pipe wall response coefficient on the water quality characteristics of water supply system, a dynamic water quality model is used to receive the distribution feature of node water quality by calculation via Monte Carlo random sampling method under the condition if their random change follows normal distribution. This paper gives the application of this algorithm to water supply system, and the results show that when all uncertain input parameters jointly take action, the uncertainty of node water quality is maximum and that water quality response coefficient is the largest parameter to affect the uncertainty of node water quality.