Acid mine drainage (AMD) generation is a serious problem from the environmental and economic perspectives, because it contains a large amount of heavy metals and because there are high costs associated with maintaining the facilities, purchasing neutralizing agents, and disposing sludges; which are required for the treatment process. In this study, changes in AMD quantity and quality for the next decades were predicted by a three series tank model in three stages, by combining the first order kinetic calculation of sulfide minerals' dissolution for two metal mines (X and Y) in Japan. Results from the AMD quality model represented the decrease of heavy metal concentrations below the effluent standard values in 30–140 years, by considering dilution and/or additional dissolution by heavy rain and snow melting, although these predicted values diverged by our previous model. However, the low correlation coefficient values (0.23–0.63) observed between the measured values of heavy metal concentration and the values calculated by our new model, mean that other chemical reactions, such as sulfate and/or carbonate mineral dissolution could greatly affect the AMD quality. In fact, there was no correlation between the metal potential calculated by our model and the real distribution of sulfide minerals at X mine. Our results therefore indicate that specific geochemical reaction and geological information should be included in the AMD quality prediction model, to estimate more accurately the fluctuation of each heavy metal concentration during different seasons.