有鑑於國內外各項大學評鑑制度與相關文獻，其評鑑指標之設計皆著重大學「內部」組織因素對學校效能的影響，以及有關大學評鑑的過程，具有群體模糊決策之性質。本研究乃嘗試在考慮外部競爭環境與模糊群體決策兩項因素之下，進行大學評鑑指標選取模式之建構。本研究主要採用：波特五力分析、德爾菲（Delphi）群體決策、與專家意見模糊整合等三項方法，進行評鑑指標選取模式之建構。該模式包括五個組成部分：大學競爭現狀分析、整合式大學競爭策略矩陣、評鑑指標資料庫、評鑑指標選取機制、與建議之評鑑指標清單等。為求所建構的模式能與現實環境配合，我們特別邀請四所新興大學之一級主管參與本研究。同時，為驗証該模式之可行性，我們並以元智大學為例進行實證分析，其驗證之方式為1)本模式所選取之評鑑指標，與2)專家憑經驗直接選取之評鑑指標，兩者進行比較。其結果顯示，本研究所建議之評鑑之評鑑指標具有系統全涵性、客觀性、與操作容易等之優點。此外，本研以不同的α-截集（意指模糊程度）來表示不同的決策環境變動性。當α-截集為0.6時，該模式選出之指標清單包括6個指標面向，其範圍涵蓋教育系統之輸入、過程、與輸出。同時根據該6個指標面向，模式總計選取36個評鑑指標。該建議指標清單所代表之意義，在本文中有進一步之闡述。

　　　　　Reviewing the past literature of performance evaluation in higher education, we found most studies paid more attention on the internal organizational factors rather than the external competitive factors. Besides, effectiveness gradually becomes a crucial issue in higher educational managerment under the limited educational resource circumstance. In order to realize how effective the university is, a fair evaluation need to be conducted. However, as we how, the evaluation committee is always a group of members and their preferences are fuzzy. The purpose of this study is tried to construct a performance indicator system by taking into account these two decision factors, which are the external competitive environment and fuzzy group decision. Three approaches are applied in this study , one is Porter's five forces analysis, the others are Delphi technique and fuzzy group decision. The proposed system is consisted of five components: the current circumstance analysis, the competitive strategic matrix, the indicator database, the selection mechanism, and the suggested indicator list. We invite the executives from four universities to participate this study. In the meantime, we conduct an empirical study in Yuan-Ze University to verify the feasibility of the proposed system. The results of empirical study revealed that the indicators elicited from the proposed system have the advantages of more systematic, objective, and accessible than those without using the proposed system. In this study, we adopt the concept of α-cut to stand for the degree of fuzziness in various decision environment. When α-cut is set to 0.6, 36 indicators were selected by the proposed system. We have a further discussion in the following regarding to those suggested indicators.