In this study, THM formation in a water distribution system is predicted by statistical analysis and neural network with variables of AWWA model. The THM formation equation by statistical analysis is determined by multiple exponential regression (Model I) and multiple linear regression (Model II). The prediction of THM formation by neural network analysis (Model III and Model IV) is determined by including organic compounds such as TOC, DOC and UV254. The result shows that THM concentration increases by 30% or less in the C-distribution system as the distance from water treatment plant increases. It is shown that the correlation coefficients of THM prediction by statistical analysis are 0.917 and 0.937 for Model I and Model II , respectively. And, the correlation coefficients of THM prediction by neural network analysis show 0.986 and 0.977 for Model III and Model IV, respectively.) From the results, it is presented that the proposed methods are reflected well the characteristics of observed THM formation. Also, it is concluded that the neural network approaches of this research can be applied effectively to the prediction of THM formation in a water distribution system.