Groundwater management is a significant task to have sustainable groundwater sources in Afghanistan especially in Kabul Basin. In this study, the groundwater level fluctuation simulation and forecast by a traditional tank model, snow tank + traditional tank model (S + traditional tank), combined tank model and snow combined tank model (SC-tank model) were compared. The variables (precipitation, groundwater level, temperature and evaporation) were utilized to simulate and forecast groundwater level fluctuations at a representative observation well (CKB1-W) in Kabul Basin from 2005 to 2013. Shuffled Complex Evolution-University of Arizona (SCE-UA) algorithm was utilized to find the best parameter for the models. Accuracy of model estimation was evaluated by coefficient of determination, Nash-Sutcliffe efficiency (NSE) coefficient and root-mean-square error (RMSE). Consequently, the SC-tank model provided the most accurate result in simulation and forecast of groundwater level fluctuations at the representative observation well in Kabul Basin. The result indicated that the SCtank model constructed in this study could be applied for groundwater management in Kabul Basin, Afghanistan.