We investigate Inventory-Location Problems where customer demand is time-sensitive, i.e., demand decays with the delivery time. The problem is to find optimal number and locations for the facilities and how much inventory should be held at each facility so that the overall revenue is maximized. In the "customer choice" version of the model customers select the closest facility; in the "direct assignment" version, customers can be assigned to any facility. These two versions have parallels in traditional and internet-based retailing, respectively. A number of structural properties for the two models are established and efficient solution approaches are suggested. Computational results show that the ability to assign customers to facilities does not confer significant advantages to internet retailers.