Benchmarking the energy efficiency of buildings is important for optimizing the efficient use of energy and reducing carbon footprint. Healthcare facilities have especially high energy usage, but the historical difficulty of collecting energy data from a relatively large number of healthcare facilities has made it challenging to develop the appropriate benchmarking system. In this paper, we seek to foster the task of benchmarking the energy efficiency of healthcare facilities using three different methods: multiple linear regression (MLR), generalized additive model (GAM), and energy performance index (EPI). The analysis was applied using a unique dataset that contained information on energy consumption and various building features for 22 large-size public hospitals managed by the Shanghai hospital development center (SHDC). Findings suggest that different benchmarking methods yield substantially different energy performance ranking results. Furthermore, a comparative analysis of the three benchmarking methods was conducted in terms of goodness-of-fit, consistency, and robustness. The results show that MLR tends to be the most consistent and robust benchmarking model, while GAM appears to have the best goodness-of-fit. The proposed methodology can assist hospital managers identify potential improvements for more efficient use of energy in healthcare facilities.