In a computer-based mail environment, digital addresses (i.e., the equivalent of telephone numbers) are sufficient to allow the mail system to perform forwarding and delivery. However, digital addresses are not easily remembered by people and some of the intelligence inherent in the system must be used to provide a more humanly engineered addressing scheme. This paper proposes a two-stage approach. In the first stage, a set of attributes (names, postal address, company affiliation, etc.) is mapped into an identifier which designates a single originator or recipient (O/R) of mail. This mapping is performed via a database query dialog and is executed prior to the posting of a mail item. The O/R identifier designates an O/R invariant to its current location and is the primary information by which the mail system decides to whom mail must be delivered. In the second stage, the mail system converts the O/R identifier into an address which is then used by the mail system for forwarding and delivery. The paper discusses the relationship between a hierarchy of authority domains and the attributes and identifiers. It also analyzes the problems associated with updating the databases that support the two mapping functions. In addition, methods are developed that allow the mail system to deliver mail items correctly even under the assumption that not all databases involved contain accurate information.