Science park has been widely recognized for its importance to the development of high-tech industries. However, as the space availability is limited, selection of firms with better efficiency and/or growth potential in specific high-tech industries to get into the science parks has become a critical issue for the Taiwan government. Accordingly, this study applies Data Envelopment Analysis (DEA), a multiple inputs–multiple outputs evaluation method, to analyze the comparative performances of the six high-tech industries currently developed at Taiwan's Hsinchu Science Park. Malmquist indices of productivity change are also used to analyze the growth potentials of the six high-tech industries. In addition, individual output/input ratio analysis is done to examine the differences on individual productivity items between the efficient and inefficient industries. The results of these analyses provide some policy implications for Taiwan and other countries facing the similar problems.