The potential implications of climate change for coastal communities and infrastructure have been a key focus of risk and vulnerability assessments. Recent experience in the Gulf Coast with Hurricane Katrina has demonstrated the vulnerability of energy infrastructure to tropical cyclones, and sea-level rise creates the potential for future escalation of vulnerability. Researchers have an extensive toolbox of models that can be applied in assessments of infrastructure vulnerability, yet few attempts have been made at model intercomparsion. To address this knowledge gap, a range of storm surge modeling approaches were compared to assess their sensitivity and specificity with respect to the identification of energy facilities along the U.S. Gulf Coast exposed to inundation from hurricane storm surge. Reports of flooding in electricity generation and oil refining facilities during Hurricane Katrina were used as a benchmark for evaluation. Those observed impacts were compared with flooding predicted by four different modeling approaches including simple models as well as models of intermediate and high complexity. Depth of inundation predicted by each modeling approach was compared with estimates of facility elevations to assess potential exposure. All modeling approaches generally had a high degree of sensitivity with respect to the identification of facilities that experienced inundation during Hurricane Katrina. However, specificity in terms of avoiding false positives was lower. This phenomenon can be attributed to uncertainties common to each modeling approach including the error associated with elevation data, lack of information on site-specific flood defenses, and dynamic processes not captured in models (e.g., levee failure). Hence, simple storm surge modeling approaches appear effective for the generic screening of potential facility exposure. However, more complex biophysical models as well as infrastructure and energy system models are needed to understand the processes by which individual storm events and local context interact to create vulnerability.