With the goal of fiscal self-sufficiency, the United States Postal Service (USPS) has embarked upon a 10-year program to modernize, and in some cases radically alter, the way it manages and processes the mail. At the heart of this effort is the goal of automating virtually all of the letter mail by 1995. This means reading, sorting, and then sequencing each mail piece to the order in which it will be delivered by the carrier with only a minimum of manual labor. In support of this goal, a series of long-term planning models has been developed to help select equipment and plan for its use at the more than 250 general mail facilities (GMF) throughout the nation. This paper reports on one of the central studies underlying this effort. Because of the size and complexity of the facility design problem, a hierarchical approach was followed. Three interrelated models were developed starting with a mixed integer linear program to derive equipment needs and initial machine schedules. The latter are post-processed in two stages to produce implementable schedules that reflect current practice. An auxiliary linear program and a heuristic were constructed for this purpose. The models and analysis conducted are demonstrated with data obtained from the Providence GMF. The results indicate savings over the 10-year planning horizon.