This work presents the method of failures forecasting in district heating systems on the basis of time series analysis.Statistical data concerning the frequency of damage in communal heat distribution networks from a ten-year time period have been used in the article. The data were subjected to statistical analysis. The methods of exponential smoothing, ARIMA, homologous period trend estimation and seasonal indicators were applied as the most adequate for this type of data. Four forecast models of damage in district heating systems were formulated and the best optimal model for the system was selected.After verification, the methods of failure prediction in district heating systems can be applied in the operating process of a heat supply system. These methods give us answers to many questions about planning of repairs, maintenance emergency service, planning of finances and purposefulness of operational control.