本文利用資料包絡分析模式（data envelopment analysis, DEA），建構以多種投入，生產多種產出（multi-inputs and multi-outputs）的大學產業之生產技術，以探討87學年至92學年我國各大學院校之技術效率。此外，利用迴歸模型解釋大學院校無效率的可能原因。 為了避免DEA估計結果發生censoring問題，本研究加入Andersen and Petersen（1993）之超效率（super-efficiency）觀念，以衡量學校的經營效率。結果顯示，以DEA模式評估，私立學校均較公立學校有效率；有設立研究所之學校優於未設立研究所的學校；而醫學院校與新改制的私立大學則表現相對高於其他類學校。上述結果均呈統計顯著性。此外，大部分的大學院校均不在最適規模上生產（或說是規模不效率）。最後，我們利用OLS迴歸分析探討各大學效率表現差異的原因，結果發現，權屬別、是否設立研究所、規模大小、每教師申請到的研究經費、生師比、學校所在地的大學家數等，是顯著影響學校效率表現不同的主要因素。

To investigate production performance of universities (including colleges) in Taiwan, we apply the data envelopment analysis (DEA) method to construct multi-input multi-output production froniter of universities in Taiwan from 1998 to 2003 academic years. In addition, regressions are used in the second stage to pinpoint factors that contribute to inefficiency among universities. To avoid the censoring problem of the DEA efficiency scores, we employ the concept of “super-efficiency” in Anderson and Petersen (1993) in our evaluation. The results show that, on average, private universities perform significantly better than public universities; universities with graduate schools perform significantly better than universities without graduate schools; medical schools and newly established private universities perform significant better than others. In addition, we also find that most universities operate inefficiently in scales. Finally, OLS regressions show that owership strutrue, with or without graduate schools, scale, research grant per teacher, student-teacher ratio, and competition, are significant determinants of technical efficiency.