Negative potassium (K) balances have been reported from grass-dominated organic systems in several European countries, and in these systems delivery of K from soil primary minerals by weathering is probably crucial to sustained productivity. Mass balances and K budgets have been made at eight grass-dominated experimental agricultural systems, one each in Scotland and Sweden and six in Norway. Where no fertiliser K was applied, the deficits obtained were in the range 35-241 kg K ha-1 per year. An estimate of the weathering rate was determined by the biogeochemical PROFILE model that varied widely between locations (3-82 kg K ha-1 per year). This paper is a first attempt to apply the steady-state PROFILE model on agricultural soils by using data from experimental sites. The Monte Carlo simulation of the weathering rate calculation showed that differences in specific surface area and soil water content strongly influenced the predicted weathering rate, which was associated with a degree of uncertainty. All sites showed a negative annual K-balance, which means that the specific contribution that mineral weathering makes to the long-term replenishment of soil exchangeable-K is important but may not be sufficient to counteract large K removal associated with high crop yields. Despite the uncertainties in the weathering rate calculation, these preliminary results are still useful in several aspects because the simulations make it possible to compare different agroecosystems with different management regimes, climate, soil properties, etc. The simulations also indicate which environmental factors are likely to be the most influential on weathering rates. Soil physical properties, such as moisture content, bulk density and surface area, appeared to be amongst the most important input parameters controlling actual weathering rates within groups of soils having similar mineralogical and chemical compositions. There is now a need for improved laboratory experimental data that better describe the kinetics of mineral weathering in order to enhance the accuracy and precision of PROFILE.