This article describes the overall development, including formulation and calibration, of linear and nonlinear multiple regression models for predicting total trihalomethane formation potential and kinetics during the chlorination of natural waters. The rationale behind each model formulation is discussed, and statistics relating to the calibration of each model are presented. The testing and attempted validation of these models are also addressed. Each model is subjected to a sensitivity analysis and a validation analysis using data derived from the literature. Includes 31 references, tables.