This paper discusses results of a simulation with the regional climate model HIRHAM for 1958-2001, driven by the ECMWF reanalysis (ERA40) data over the Arctic domain. The aim is to analyze the ability of the model to capture certain features of climate extremes derived from daily mean, maximum and minimum temperatures. For this purpose, a range of climate indices (frost days, cold and warm spell days, growing degree days and growing season length) was calculated from the model output as well as from ERA40 data and region-specific station data for Eastern and Western Russian Arctic for comparison. It is demonstrated that the model captures the main features in the spatial distribution and temporal development of most indices well. Though systematic deviations in the seasonal means occur in various indices (frost days, growing degree days), variability and trends are well reproduced. Seasonal mean patterns in frost days are reproduced best, though the model persistently calculates too many frost days. Seasonal means of cold and warm spell days are reproduced without systematic biases, though deviations occur in summer for cold spells and in spring and summer for warm spells due to an early spring warming in the regional climate model and a low variability of the daily maximum temperature over sea ice.