With cities accounting for approximately two thirds of the global demand for energy, there is significant scope to optimize energy usage of cities, in particular by improving the use of the built form. Large non-domestic buildings are increasingly the focus of attention, due to their substantial demands and associated environmental impacts such as CO2 emissions. Various approaches have been adopted to address building energy efficiency, with more recent studies relating consumption patterns to human occupancy. This paper proposes a new method to measure activity, using WiFi connections as a proxy for human occupancy. Data on the number of WiFi connections and energy consumption (electricity. steam and chilled water) were compared for two buildings within the Massachusetts Institute of Technology's campus. The results of the study demonstrate: the operation of the heating, ventilation and air conditioning (HVAC) systems adhered more closely to factors other than occupancy i.e. external temperature, whilst a small part of the electricity levels did correlate with the occupancy. In order to present possible solutions to address the disconnect between the HVAC system and occupancy levels, this paper identifies future steps that could begin to improve energy usage. (C) 2011 Elsevier B.V. All rights reserved.