To evaluate the sensitivity of mountain permafrost to atmospheric forcing, the dominant meteorological variables such as temperature, precipitation and timing and duration of snow cover have to be considered. Simulations with a one-dimensional coupled heat and mass transfer model (CoupModel) are used to investigate the interactions between the atmosphere and the ground focusing on ground temperature evolution and the temporal variability of the depth of the unfrozen top layer in summer (active layer depth). Idealised and observed atmospheric forcing data sets are used to determine the meteorological conditions, which show the largest impact on the permafrost regime. Borehole temperature and energy balance data from the permafrost station Schilthorn (2900 m asl, Berner Oberland) are used for verification. The results for the Schilthorn site show the largest impact due to summer temperatures changes during the snow free period and to a lesser extent winter precipitation which influence the duration of the snow cover. Similarly important is the timing of the first snow event in autumn which leads to a sufficiently large snow cover to isolate the ground from atmospheric forcing. Simulations with different data sets from Regional Climate Model (RCM) simulations derived from an ensemble of models and scenarios show that the differences in changes of active layer depth between different RCMs are on the same order than between different scenarios.