In this study, Data Envelopment Analysis is used to measure relative efficiencies of academic departments of an engineering college. Input and output criteria are determined and measured utilizing the academic personnel performance measurement scheme of the College. New measures are developed to compare departments of different disciplines. The discriminating power of the measurement models is improved by restricting criteria weights directly and by using a multiple criteria decision making approach. Sensitivity of the results to exclusion of a criterion and changes in relative importance of the criteria is analyzed. Evaluation of the results jointly by the college administration and the departments is explained for a successful efficiency improvement mechanism.