This paper presents part of the results of a large- scale, long-term experimental research conducted at the Faculty of Civil Engineering and Architecture Osijek. Among other research goals, this research aims at further development and improvement of a relatively new method used for the measurement of ther- mal transmittance of walls (U-value) in literature, often called temperature-based method (TBM). This research also partially overlaps with other researches carried out at the Faculty of Economics in Osijek, where the main research goals were development of machine learning and neural network models for predicting energy consumption in buildings, which will reduce the energy performance gap between design and actual energy needs. Building thermal performance as a whole can be quantified by the heat loss coefficient (HLC) or the total heat loss (THL). Experimental research presented in this paper was conducted by using a built test chamber in a laboratory, and the research lasted for 40 days. This is an innovative element of this research, since the test chamber is built inside a laboratory where external weather conditions are simulated by omitting the negative influence of wind, precipitation, and solar radiation on the experimental results. The actual heating energy consumption by the test chamber was recorded daily for 40 days during the winter season, together with internal and external temperatures, relative humidity (RH), U-values of walls, and wind speed. Chamber airtightness was measured at the beginning of the experiment. These measurements made it possible to perform the Co- heating test. This test is used to calculate the total heat loss of a building, both fabric and ventilation loss. Parallel with the Co-heating test, the design heating energy need of the test chamber was determined by calculating the heat loss coefficient and the total heat loss. Actual and design values of heat loss coefficient and total heat loss were used to characterize the energy performance gap. Energy performance gap in this study was found to be between −40% and 13%. Research results indicate the variables affecting the actual and design values of heat losses significantly. Presented results provide guidance for more accurate determination of actual energy consumption in buildings, and therefore help in the reduction of the energy performance gap.