The present paper deals with the heating potential of a single buried pipe using real climatic data. The use of buried pipes in buildings for heating and cooling purposes has gained increasing acceptance in recent years. The system's heating potential was calculated using an accurate, dynamic, deterministic, numerical model. Multiyear ambient air and soil climatic data for the city of Athens have been used as inputs to the deterministic model and the results were compared. Furthermore, a neural network approach was used for estimating the thermal performance of the system in heating for the city of Athens. Moreover, the influence of several climatic parameters used as inputs to the neural model such as the ambient air temperature, the ground temperature and the relative humidity is investigated and analysed. (C) 2002 Elsevier Science Ltd. All rights reserved.