The growing attention towards the building energy efficiency has significantly increased the number of studies in this field. Nevertheless, they are still based on outdated reference models which adopt internal loads derived from old studies, ignoring recent technological and functional changes. The main aim of this paper is to assess the incidence of 8 input variables – 1) occupancy, 2) lighting power density, 3) equipment power density, 4) outdoor air rate, 5) infiltration rate, 6) minimum and 7) maximum supply temperature, and 8) daylighting setpoint – on thermal consumptions to identify the most influential variables and provide updated reference values for an office case study. To that end, 25 cities were considered to conduct a local one-at-time sensitivity analysis simulating, in each location, 11 alternatives for each variable. The resulting 2200 consumption values were statistically analyzed and highlighted a strong dependency on the first four variables (1–4). These variables were further investigated to define two updated sets of reference values, for new construction and refurbished buildings. These sets were proposed for a representative office building located in Milan which could be used as reference in multi-contexts studies when no certain/measured data are available. Finally, the consumption differences between starting and proposed sets were compared highlighting the importance of these input values.