This paper reports on various options for estimating room occupancy levels in an office building from the results of testing, and aims to find an efficient way to represent this occupancy, optimizing accuracy, cost effectiveness and the intrusiveness of measuring, which is very useful for commercial applications. CO2 concentration and computer electricity consumption were considered alternative indicators of occupancy level. Advantages and drawbacks of both indicators are discussed. In both cases, ranges when the room was positively empty and when it was positively occupied were identified. This study is based on histograms that represent indicator measurements in each of those ranges during the study period. Afterwards, the ranges identified were used to find room occupancy patterns. Information previously available about the rooms, such as typical work hours, and lunch hours, holidays, weekends, etc., when rooms are empty were used to validate the results. Data were from a seven-year series acquired while the building was in regular use. The first analysis considers all of these available data making results robust. Accuracy with shorter experimental test periods was also studied. (C) 2017 Elsevier Ltd. All rights reserved.