We have developed a model-based soil organic carbon (SOC) inventory for mineral soils under permanent grassland and cropland in Switzerland to be used for National Greenhouse gas reporting under the UNFCCC. The inventory system is based on the soil carbon (C) model RothC and depends management of the 19 most important crops and 6 grassland categories. An allometric equation is used to derive the amount of plant C inputs to the soil based on measured yields. Meteorological data were derived from the Swiss meteorological service. The clay content of the soil was roughly estimated based on a soil suitability map. To calculate initial SOC stocks we used an approach that relates SOC stocks to clay content, altitude and land-use type. The size of the different C pools in RothC was estimated using a pedo-transfer function, which proved to be a good alternative to the estimation with a spin-up. For national-scale simulations the country is stratified into 24 homogeneous regions with similar climatic conditions and agricultural production types (characterized e.g. by the steepness of slopes or accessibility). For each of these 24 strata a simulation was run for 25 different crop and grassland types, for 10 different soil clay classes. The final SOC time series for each stratum was calculated as an area-weighted average for each combination of crop and soil type. This was found to be an acceptable alternative to simulating real crop rotations. The system is dynamic, capturing inter-annual variability in SOC changes due to, for example, meteorological conditions, cultivated crop types and herd sizes. Furthermore, it is flexible allowing for continuous improvements as well as the representation of some future changes in management. Finally, the inventory system can also serve as a tool for sensitivity analysis or to explore specific GHG mitigation options that increase SOC stocks. In the National inventory report, the results are aggregated for three elevation zones and are reported separately for permanent grassland and for cropland. An initial uncertainty analysis based on Monte Carlo simulations considering uncertainty in the input parameters revealed however that the mean relative uncertainty of year-to-year SOC stock changes is greater than 100 % indicating that both land-use types can be considered as C neutral.