Residual chlorine concentration is an indirect indicator of the quality of distributed water. Originators and managers of networks use this indicator in the dimensioning process and in their practical decisions. The model used for this study is the quality module of the water distribution system simulation software PORTEAU. The parameters required for the calculation of the concentrations are the kinetic constants of reaction of the different pipes. The lack of precise knowledge concerning these constants prevents us from calculating them explicitly, so we must proceed to a calibration. Measurements are then necessary to guide the choice of the parameter values. This calibration depends on the place of the measurements. Different methods transposed from the hydraulic field are tested to get the best alternative to the potential points of measurement. This choice is based on the minimization of the error made in the estimation of the parameters. After analysis and comparison, the methods suggested by Bush (1998) and Piller (1999) are adopted, programmed and applied to a study network. They both use a sensitivity matrix, whose coefficients are built, for our study, on a theoretical study carried out on simple elements, widened to the whole network. The method proposed here constitutes a means for approaching the derivatives composing the real coefficients of sensitivity. These matrices are then treated to give a classification of the possible combinations of measurement points and thus provide the best choice. Once the points have been chosen, automatic calibration can be set up with a Levenberg Marquardt algorithm.