This study presented a model of constructing expert system knowledge base for an undergraduate computer networking class. This model included (1) identifying a suitable problem, (2) defining the problem domain, (3) specifying goals or solutions, (4) specifying problem attributes and values, (5) generating rules and examples, and (6) selecting the right tool. In addition, the effectiveness of this problem-based learning approach was verified by an experimental study. Results indicated that students creating their own expert decision-making system based on IF-THEN rules achieved significantly higher scores than students guided by an expert system already developed by experts (F=5.042, p< .05) when they were solving story problems. The same instructional technology can be applied to other disciplines that focus on teaching engineering problem solving.