If renewable energy sources are to be successfully utilised within the built environment then accurate simulation tools have an important role to play. This paper addresses a very specific part of the overall simulation process: that of earth-contact heat transfer which can be the cause of significant modelling errors Using a three-dimensional numerical model, simulation results are compared with experimental data for a test room. It has been shown that for this case, neglecting climatic conditions (e.g. snow and rain) causes discrepancies between measured and calculated temperatures. A sensitivity study has also shown that thermophysical properties (like the soil thermal conductivity) have to be accurately known.