Large-scale disasters cause enormous damage to people living in the affected areas. Providing relief quickly to the affected is a critical issue in recovering the effects of a disaster. Pre-disaster planning has an important role on reducing the arrival time of relief items to the affected areas and efficiently allocating them. In this study, a mixed integer programming model is proposed in order to pre-position warehouses throughout a potential affected area and determine the amount of relief items to be held in those warehouses. Time between the strike of the disaster and arrival of relief items at the affected areas is aimed to be minimized. In addition, using probabilistic constraints, the model ensures that relief items arrive at affected areas within a certain time window with certain reliability. Considering instable fault lines on which Istanbul is located, the proposed model is applied to the Istanbul case for pre-positioning warehouses a priori to the possible expected large-scale earthquake.