To support the ASHRAE Building Energy Quotient (bEQ) rating program, ASHRAE TRP-I711 proposed a methodology to reconcile the differences between the empirical and modeled baselines of existing and new US commercial buildings. As a part of this research, this paper details and applies a new methodology to systematically identify sensitive inputs of building energy models. First, the methodology is introduced to identify possibly sensitive model inputs, determine the uncertainties of these inputs, and conduct sensitivity analysis to select sensitive model inputs. Then, to demonstrate an application of the methodology, we evaluate the sensitivity of energy model inputs for medium office buildings. We develop 30 prototypical building models that represent pre-1980 and post-1980 medium offices in the 15 ASHRAE climate zones. Fourteen model inputs, which are identified as being potentially sensitive, are selected for each prototypical building model. Sensitivity analyses are performed as desm'bed in the methodology. The results indicate that the lighting power density and plug load density are the two most sensitive model inputs. Meanwhile, the envelope is more sensitive in cold climates than in hot climates, and shading devices are more effective in hot climates.