In high mountain areas, permafrost is important because it influences the occurrence of natural hazards, be- cause it has to be considered in construction practices, and because it is sensitive to climate change. The assessment of its distribution and evolution is challenging because of highly variable conditions at and below the surface, steep topog- raphy and varying climatic conditions. This paper presents a systematic investigation of effects of topography and cli- mate variability that are important for subsurface temper- atures in Alpine bedrock permafrost. We studied the ef- fects of both, past and projected future ground surface tem- perature variations on the basis of numerical experimenta- tion with simplified mountain topography in order to demon- strate the principal effects. The modeling approach applied combines a distributed surface energy balance model and a three-dimensional subsurface heat conduction scheme. Re- sults show that the past climate variations that essentially in- fluence present-day permafrost temperatures at depth of the idealized mountains are the last glacial period and the ma- jor fluctuations in the past millennium. Transient effects from projected future warming, however, are likely larger than those from past climate conditions because larger tem- perature changes at the surface occur in shorter time peri- ods. We further demonstrate the accelerating influence of multi-lateral warming in steep and complex topography for a temperature signal entering the subsurface as compared to the situation in flat areas. The effects of varying and un- certain material properties (i.e., thermal properties, porosity, and freezing characteristics) on the subsurface temperature field were examined in sensitivity studies. A considerable influence of latent heat due to water in low-porosity bedrock was only shown for simulations over time periods of decades to centuries. At the end, the model was applied to the topographic setting of the Matterhorn (Switzerland). Results from idealized geometries are compared to this first example of real topography, and possibilities as well as limitations of the model application are discussed.