Contemporary CASE tools contain a certain types of knowledge to provide their users with a limited intelligent support for a development methodology. For more intelligent support, various types of knowledge should ne incorporated into CASE. The knowledge can be categorized as structural and behavioral knowledge and operational knowledge. The operational knowledge is again classified as knowledge on the methodology constraint, constraint checking time, way of constraint check, and informative feedbacks of a CASE tool. A metasystem should capture these of knowledge in it``s metamodel to generate a practival CASE tool. Contemporary metamodels, however, can represent only structural and behavioral knowledge because they are based on Entity-Relationship (ER) or extended ER data model. This thesis proposes and implements a representation scheme of the operational knowledge in a metamodel called RBM (Rule Based Metamodel). In the new metamodel, operational knowledge is represented as rules. Each rule is defined in an object/relationship type definition or as a constraint type. A rule consists of three parts : event part, condition part, and action part. Event part defines the triggering time (condition checking time) of the rule. Condition part identifies methodology constraints associated with the rule. Action part is used to specify informative feedback knowledge which is given to users when they violate the rules of the CASE tool supporting methodology. A new metasystem based on the RBM is implemented in C and inference engine of the CLIPS under PC environment.