Millions of deaths occur annually in developing countries due to diarrheal diseases caused by poor water quality. Household water treatment and safe storage interventions have been shown to improve water quality and reduce diarrheal disease incidence in developing countries. Chlorination is an intervention that is widely used to treat contaminated drinking water. Benefits of chlorination include scalability and low cost; a drawback is ensuring quality control over time. The objective of the current study was to develop a predictive model of chlorine concentrations in drinking water. The predictive model consists of single and multi-compartments described by systems of linear differential equations. Assumptions of the model are: (i) the volume of the container is fixed; and (ii) there is no spatial dependence of chlorine in the water system. The model code is written in Berkeley Madonna Software (Berkeley, CA). Experimental data from Vestergaard Frandsen's Water Laboratory (Hanoi, Vietnam) was used to calibrate the model. Laboratory experiments were conducted to measure the decrease of residual chlorine concentration as a function of time in aging water. The tests were conducted using water with different chlorine concentrations, organic content and turbidity levels. Simple compartment models fit the laboratory data well. Simulations of chlorine residual levels were conducted using Monte Carlo techniques, varying the initial concentration of chlorine, rate constants, organic content, turbidity levels, and chlorine interventions. This predictive model may be used by chlorine-based water treatment programs to simulate chlorine concentrations for a variety of water use patterns, and provide quality control of the program.