A simplified tool, called ConsoClim, has been developed for estimating energy consumption of air conditioned buildings and for comparing HVAC and building solutions. Its main aim is to be used at the early stage of the design process. It means that algorithms have been developed to be used with minimum and quite simple inputs that are available in manufacturers' catalogues or can be fixed at typical default values. The aim is to check that building design options and HVAC system choices lead together to an efficient global energy performance. A prototype of the software package using ConsoClim algorithm library has been achieved. First comparisons of the results with in situ measurement are reported in this paper. As a first step, the comparison deals only with heating and cooling building demand. Thus, an existing building connected to an urban hot and chilled water distribution network is used. Its energy demand is read at the heat exchanger and full year monthly data file is available. The building is described. But there is no measurement of the internal gains. Electrical appliances are estimated by monthly electricity consumption. Moreover, little specific information is available about occupancy. Then, its scenarios are based on typical office building hours. The uncertainty about some input values is therefore very large. Uncertainties about each input are estimated. Heating and cooling demands are thus estimated by ConsoClim simulations and compared, on a monthly basis, to the measured delivered energy. Results give conclusive elements about validation and show how energy calculation is sensitive to the choice of conventional input data.