A review of research on the life cycle costs of buildings shows a lack of empirical research linking building characteristics to life cycle costs. Particular emphasis is placed on the lack of useful, reliable and consistent data needed to estimate the total life cycle costs of buildings. The general objectives of this research are to collect and analyze data on energy performance, energy consumption for space heating and heating energy costs of school buildings in the Federation of Bosnia and Herzegovina (FBiH) to investigate the impact of individual characteristics (variables) of buildings on heating energy costs. The special goal of this research is to develop new mathematical models for faster and simpler estimation of heating energy costs of school buildings, which will be based on the energy performance of school buildings in FBiH. For this research, data from the documents of detailed energy audits of school buildings in the FBiH were used, which contain a large amount of data on energy characteristics, energy consumption for space heating and heating energy costs. In the first step, a statistical analysis of the energy performance of school buildings in the FBiH in relation to construction periods was performed based on the data collected from the sample. By comparing the actual (measured) and predicted (calculated) delivered energy for space heating, it was determined that the actual consumption is on average about 65% of the predicted. This conclusion is very important because the actual energy consumption for space heating does not reflect the needs, and in the development of mathematical models for estimating the heating energy costs, the calculated values of energy for space heating were used. In the second step, the analysis of independent variables was performed and the separation of the basic data set into the training set and the set used for the validation of regression models was performed. The most common statistical models for estimating the energy performance of buildings are regression models that link energy consumption and heating energy costs to one or more variables. Multiple linear regression models have been developed in advance using the stepwise regression method, and some of the statistical methods have been used to estimate them. The criterion of representativeness and the criterion of simplicity were set for the evaluation and ranking of the developed regression models. The developed regression models confirmed the hypothesis of this research that independent variables or performance of school buildings can be used to estimate dependent variables, ie the heating energy costs of school buildings in the FBiH.