Abstract: Data Envelopment Analysis (DEA) is a widely used non-parametric technique for measuring technical efficiency of organizations that use multiple inputs to produce multiple outputs, when prices of inputs or outputs are not known. This dissertation uses DEA to explore the technical efficiency of doctoral programs in the United States as well as institutions of higher education in China. Using data from the 1995 report of the National Research Council, the U.S. study examines the extent and possible sources of relative inefficiency in 1665 Ph.D. programs across 15 academic fields. Most programs in any given comparison appear not to be relatively efficient. However, the determinants of technical efficiency remain largely unknown: no consistent correlations were found over multiple fields between DEA efficiency scores and such factors as the ratio of graduate students to faculty, the median years obtaining a degree and reputation ranking. One pattern that did emerge was that, in many fields, programs in public institutions have lower efficiency scores than their private counterparts even after controlling for several potential confounding factors. The Chinese study examines technical efficiency of 52 universities and 5 colleges within one university. Based on the findings from the DEA measurements, some policy implications are suggested for policy makers in the country where the higher education system is undergoing dramatic restructuring. This study gives special emphasis to testing the sensitivity of DEA to different input/output variable selections, and constructs confidence intervals for the DEA scores using jackknife and bootstrapping techniques. The ranking of programs based on DEA scores does vary according to the inputs and outputs included in the model, but certain institutions are found consistently to have programs that appear efficient relative to programs in the same field at other institutions.