Abstract: As expert system technology spreads,the need for verification of knowledge in expert systems assumes greater importance.This paper addressed the issues involved in demonstrating an expert system based on if-then rules and frames to be free from errors.Knowledge representation and contents of knowledge base are briefly discussed.Errors normally encountered in knowledge base such as redundancy,conflicts,circularity and incompleteness of coverage are outlined and analyzed systematically.In addition,their influences on system performance are also discussed.