For modelling permafrost and its response to a changing climate the soil thermal properties are crucial parameters. Those properties include the thermal conductivity, the thermal diffusivity and the heat capacity of the ground. However, in the vast Arctic region data on thermal soil properties are only sparsely available. Therefore scientists try to simulate processes that are not measurable or require much effort. The results show that the model, which fittest least to measured data, is the one developed by Johansen (1975). Nevertheless it is the most widely used in the land surface schemes of large-scale climate models since the input parameters, i.e. the organic content and the grain size distribution of the soil, are comparatively easy to obtain. On the other hand output data from this model differ by a factor of 1.2 or 20% from the measured values which means that the ground heat flux used for climate scenarios has an offset of that range to reality. A better solution would be to use the model of de Vries (1952). It requires more detailed knowledge about the soil composition and the several constituents. Requiring spatial information about these parameters is necessary when trying to incorporate the model into regional and global simulations but poses a challenging tasks for large regions and especially the remote Arctic areas. Using a simple but inadequate model like Johansen, however, largely biases modeling results of ground heat flux which is a crucial parameter in projections of future Arctic climate change. Correct parametrization of local soil parameters as implemented in the de Vries model are essential to adequately represent soil spatial heterogeneity in permafrost terrain and associated processes.