Abstarct: The environment is a key pillar to the sustainable development. The development of industries, the growth of the economy and the increase in the amount of industrial production lead to the production of high amounts of waste materials, worthless and harmful side products that are the source of various environmental problems. In this research, after field sampling (on the surface and in the depth) from the tailing of Mouteh gold mine, and laboratory measurements of the existing cyanide in the samples, the modeling of the cyanide distribution in the tailing is performed. For modeling, new methods of the machine learning and the geostatistics are used in MATLAB software. Therefore, the methods of Support Vector Machines, Random Forests, Linear Discriminant Analysis and AdaBoost were used for modeling, which AdaBoost method provides the highest accuracy in estimation. Geostatistics is used to cover the weakness of these methods in identification of isotropy/anisotropy of the environment and to calculate empirical errors. According to the results of geostatistical studies, the concentration of cyanide in the tailing is homogeneous and the range of variogram is 8.3 meter. This range is in agree with the number of selected samples for the estimation. The obtained value from the nugget shows a negligible error of sampling, preparation and analyzing of the samples. Also, the conceptual model of evolution and transfer of cyanide in the tailing of Mouteh gold mine is presented in three dimensions and considering the governing physicochemical mechanism. In the last step, with estimation of total amount of existing cyanide in the tailing, the cost of purification of tailing is calculated and the ways to making profit from this tailing are investigated.