Oil reserves can be estimated by numerical simulations in a waterflooded production system. For accurate reserves estimations, reservoir models should be calibrated by production history data with optimization schemes, which typically ensure better matches on the production data and permeability distribution. More reliable estimations with uncertainty could be conducted through stochastic estimations after history matching. The developed ensemble Kalman filter history matching technique can overcome the non-uniqueness problem and efficiently estimate oil reserves for decision-making with uncertainties.