A method based on a genetic algorithm (GA) combined with Levenberg-Marquardt (LM) method applicable to the population balance model for coalescence parameter estimation in a liquid-liquid biphasic system is presented. The toluene/water system in a rotating disk contactor was taken as an example. Estimation methods for such a problem are often based on deterministic optimization models that are rather instable and divergent around a local minimum. To overcome these limitations, the present study involves the introduction of a semi-stochastic method that is able to provide at first the estimation of coalescence parameters from the GA based on an inverse approach, exploiting the principle of GA. The LM algorithm was applied to ensure that the results are not restricted to a local minimum.