Knowledge acquistion is known to be a critical bottleneck for building expert systems. One of the most difficult problems of knowledge acquisition is to integrate domain knowledge from multiple experts, especially when inconsistencies and conflicts need to be resolved. In this paper, a knowledge acquisition method, MERGE, is proposed to cope with this problem. Some experimental results of a practical application, Differential Diagnosis of Common Causes of Inflamed Eye, have shown that MERGE is able to enrich the elicited knowledge and improve the performance of the construted expert systems by integrating expertise from multiple experts.