At the beginning of the process of calibration of a water quality model, using the computational tool Qual2kw that includes a genetic algorithm as a mathematical tool for calibration, it is necessary to introduce some operators for the start of the calibration process that seeks the best combination of constants that represent the reality of the current in terms of water quality. In this work are made general recommendations on three operators that uses genetic algorithm: the seed used, the number of generations and the numberof populations; mainly the latter two are important because they involve partners computational times, since a combination that creates many runs could not present significant variations in the total adjustment of the model, so that a combination “optimal” could give good solutions in reasonable time. This study found that indeed there are points where the improvement in the quality of adjustment does not increase more than 5% variation in the value obtained by the function of error, so it is possible to recommend certain values for use by the modeler at the time of use this tool.