ABSTRACT. Chlorination is the most widely practiced form of disinfection in the US. However, there is concern that the disinfection by-products (DBPs) formed during chlorination might be carcinogenic. Because of this increasing concern in water supply systems, there is a need for models that can be used to predict chlorine residuals and optimize the disinfection practices. This paper presents a chlorine decay model based on the possible chlorine decay mechanisms. To evaluate this model, four raw surface and alum treated waters (Chester, Garden City, Maysville, and Lake Vandalia) were used. The chlorine residual at the end of the study period was maintained at the same concentration to avoid effects of chlorine concentration difference. Results show that this model predicts the chlorine residual extremely well, consistently yielding correlation coefficients greater than 0.98. Alum treatment substantially increased the fraction of rapidly reacting functional groups by 24% and decreased the specific chlorine demand (SCD) by an average of 14.4%. Therefore, alum coagulation processes may preferentially remove natural organic matter (NOM) having a slower reaction rate (with chlorine), higher specific chlorine demand, and higher chlorinated DBPs production.