Though there have been numerous researches applying expert system to stock investment domain, there were not sufficient consideration about dynamic aspects. Therefore, due to the lack of consideration about dynamic aspects, many systems have faced some problems : speed problem, knowledge maintenance problem etc. In this research, dynamic management scheme of expert system for stock investment is attempted. The proposed architecture accomplishes the dynamic management of knowledge-based stock investment advisory system in dynamic environment. For the purpose of satisfying this necessity, this research proposes four schemes : (1) categorization and visualization of knowledge, (2) credibility adjustment, (3) knowledge query, (4) knowledge compilation. Inefficiency of knowledge maintenance due to knowledge compilation can be overcome as partial compilation. The other issues in this research are the reflection of dynamic situation in the process of inference, the dynamic integration of expert``s knowledge base with investor``s knowledge base, and the support for multiple purpose according to investor``s characteristics, request. To realize the dynamic management of knowledge-based stock investment advisory system, the prototype has been implemented using Common LISP in IBM-PC 386 environment.