Chlorination is the most widely used method for disinfection of drinking water, but there are concerns about the formation of by-product, such as trihalomethanes (THMs), since the chronic exposure to them may pose risks to human health. For these reasons regulations fix maximum acceptable THMs levels throughout distribution networks, so it is very important to be able to correctly reproduce their formation. In literature many models for predicting THMs formation, based on empirical relationships, have been proposed. In this work 18 literature empirical models have been selected for investigating their reliability in real situations through the application to a real water supply system in Southern Lazio (Italy). The comparison of the performances of such models furnishes interesting observations, indicating that the formula, developed using data collected through a water system, results more suitable in reproducing THMs formation for the presented case-study