In this paper we present a model that can reconstruct water table position and soil temperature profiles to 3 m depth in boreal mixed mire systems using the readily available climate data on air temperature and precipitation as driving variables. The model simulates complete, multiple annual cycles including winter conditions and freeze-thaw processes. The major requisite for an accurate description of the soil heat flux in a mire is an accurate description of the water content of the profile because of the high porosity of the soil and the thermal properties of water. The soil moisture profile in this model is described as a function of water table position derived from empirical published data of soil moisture profile measurements at different water table positions. The parameters of the model were set and tested for a lawn community, one of the most dominant plant communities occurring in boreal mixed mires. The model is optimized for the period 1995--1997 at the mixed mire Deger¿ Stormyr and was validated for the period 1997--1998 at the same location and for 1991--1993 at Stor¿myran, 60 km away. The mean deviation of simulated to measured soil temperature from 10 cm to 3 m depth in the validation data sets was <1.6 ¿C with a maximal standard deviation of 1.8 ¿C. A response analysis of the amount of winter precipitation showed the influence of snow cover depth on soil frost fluctuations. A simulated precipitation corresponding to 25% of the measured precipitation during winter 1995--1996, with all other factors unchanged, prolonged the period of frozen ground in the model from June 8 to August 1. With a doubling of the winter precipitation the modeled period of frozen ground lasted only until May 6. ¿ 1999 American Geophysical Union