One of the advantages of current database systems is the capability to enforce global integrity constraints on a large amount of data. Input of contradicting data will be rejected by database systems in order to maintain correctness. On the other hand, in medical information systems it may be necessary to realize two or more databases in one system, where there are some controlled contradictions among these databases. For example, if a doctor wants to hide the real disease name from the patient in critical condition, the database viewed by the patient should be different from the real database, although each of these databases should be conflict free and large amounts of data are shared by both. This kind of problem was not discussed for commercial business-oriented databases. Data sharing and data security are important functions required for medical information systems. There are, however, cases when we need to show non-real data to some users. Security mechanisms usually prevent a user from retrieving critical data. If a request for retrieval of some data is rejected by the system, a user may find there is something secret being kept from him. In order to cope with these situations, we introduce the POSTGRES database system. POSTGRES is a generalized relational database system developed at the University of California. The form of POSTGRES rule is as follows: On event (To) object WHERE POSTQUEL-qualification Then Do [instead] POSTQUEL-command(s) The POSTGRES rule shows that event is retrieve, replace, delete, append, new (i.e., replace or append) or old (i.e., delete or replace). The concept of objects is introduced in object-oriented databases. In relational database systems, an object corresponds to each data value, an attribute, or a relation. The optional keyword "instead" indicates that the action indicated by POSTQUEL-command(s) is to be performed instead of the action which caused the rule to activate. This keyword plays very important role for our purpose. By preparing a standard story which is consistent with the patient's condition, the doctor can make a smooth explanation to the patient showing the stories instead of the critical data. So, if medical personnel and system designers prepare the standard stories or data for the explanation and store such data in the system, the users of medical information systems (doctors) can take the benefit of the data replacement. For such a purpose, medical personnel and system designers classify the patients' data according to their age, sex, occupational history, personal history, and diseases. Standard story or data should be prepared for each user class in advance. Using POSTGRES rule system in medical information systems, doctors can make a smooth explanation of a patientUs condition in serious cases, when hiding real data is required. The validity of POSTGRES rule system is proved where data exchange is needed for data protection. For example, when a doctor writes a prescription of a placebo through the ordering system, POSTGRES rule system will identify its existence. We believe that POSTGRES rule system can be applied to many fields in medical data processing.