It is widely believed that smart metering will lead to high power savings. Those practices rely on Non intrusive Appliance Load Monitoring (NIALM) methodologies. In this work an extension of NIALM to thermal loads is proposed: Non-intrusive Thermal Load Monitoring (NITLM). NIALM and NITLM share the same key point: a good model to calibrate loads. Thermal loads calibration is, at present, a task far from trivial. This paper addresses the set of measurements required to calibrate Building Energy Simulation (BES) models. This set is found by sensitivity analysis. Weather variables together with indoors air and surface temperatures form a complete set of non-intrusive measurements for free running BES calibration. Adding non-intrusive measurements such as indoors CO2 concentration and door/window is required to validate BES under regular operation conditions. The proposed NITLM methodology validity is verified on two real-world buildings in different locations. From a complete year of measurements a BES model is validated and calibrated in both buildings for heating and cooling periods. Validity of the model is cross-checked by identifying thermal properties of the fabric and mechanical air exchange rate. Some NITLM applications are addressed and future work to extend NITLM applicability in real cases is pointed out. As example, simulation design hypothesis are checked for one building. It is shown that thermal load estimation can deviate up to three times depending on the simulation hypothesis selected. (C) 2017 Elsevier Ltd. All rights reserved.