本研究旨在利用某商業銀行78年初至83年6月之201筆授信案件作為研究對象，觀察個別客戶的授信風險，並試圖藉之進行授信客戶的甄選。實證結果發現，銀行授信客戶的甄選，首重與授信政策有關之配合度，對於客戶信用評等的重視則次之；就準則層而言，則以擔保品最受重視，往來情況次之，而財務結構與經營效率則最不被看重；至於用以反映評估指標的第四層級，則以擔保品的內容及擔保成數最受重視；客戶是否曾有違約紀錄，則居於次要地位。整體而言，層級分析模型對貸款客戶的評分，其準確性至少維持在八至九成左右，有時則甚至高達95％；新銀行之開放設立，確可導致個別銀行承擔較高的授信風險，值得金融主管機關加以正視。

This study is aimed at observing the loan risks on corporate clients and attempting to base the selection of loan clients on AHP, logit and discriminant analysis, using the 201 loans released by international commercial banks as the objects of study. Empirical results indicate that in the selection of loan clients, the bank stresses: first, conformity with the loan policy and then credit ratings of the clients; from the perspective of guiding principles, securities are the most valued, business association is the next, and financial structure and management efficiency are the least from the perspective of the fourth tier which reflects the evaluation indicators, contents and percentages of securities are most important and whether clients have ever defaulted is secondary. On the whole, an accuracy of approximately 80% to 90%, or in some cases even up to 95%, could be achieved when using the analytic hierarchy process to evaluate loan clients; the openness toward the establishment of new banks indeed led to higher loan risks from individual banks, which deserves attention of the financial authorities.