To understand the contribution of nonexchangeable potassium to plant nutrition the release and fixation of K was studied on a range of soil samples of varying K status, obtained from a long-term fertilizer trial. The soil samples were either air-dried or not prior to use so as to determine the influence of pretreatment on release and fixation of K. Sorption and desorption of K were measured by following the changes in solution K and exchangeable K, and calculating the relative contributions of exchangeable and nonexchangeable K to K dynamics in the soil-solution systems. The change in the amount of nonexchangeable K is proportional to the difference between the initially imposed concentration and the concentration for which neither sorption nor desorption occurred. Whether release or fixation occurs depends on the sign of this difference. Soils containing much K release more K than those containing little K at the same initial or final concentration in solution. The threshold exchangeable potassium and the threshold K concentration for which release becomes large increase when K content increases in the soil. For air-dried soils K-enriched soils fixed less K than those containing little K. But when kept moist the soils fixed no K. The threshold exchangeable potassium of release and the minimal exchangeable potassium, i.e. that part of exchangeable potassium extracted by ammonium acetate but which is not in exchange equilibrium with Ca, are useful criteria to assess the meaning of the exchangeable potassium when used as a fertility indicator.