The issue of sustainable soil K management has partly been ignored during the last decades when the potential environmental impact from agricultural use of nitrogen and phosphorus has been considered a more important problem. It has become evident that mixed systems, especially under organic farming practices, commonly suffer from negative K field balances, which may deplete the soil in exchangeable K if release of K from soil minerals does not replace the amount of K that is harvested. In this study, release and fixation rates of K (kg ha(-1) yr(-1)) were estimated in five long-term field experiments located on varying parent materials in South and Central Sweden, each having an array of K fertilizer application rates. The calculations were based on (i) accumulated K inputs and outputs from the soil during ca 30 yr, and on (ii) changes in topsoil and subsoil exchangeable K. The estimates ranged from ca 65 +/- 7 kg ha(-1) yr(-1) of K released from soil reserves, to 85 +/- 10 kg ha(-1) yr(-1) of K trapped by fixation. Release and fixation rates were highly dependent on the soil K balance, confirming that these are reversible processes that depend on plant uptake and fertilizer input. When receiving no K fertilizer, a loamy sand soil released 85 +/- 10 kg ha(-1) yr(-1), which was less than the annual net K removal from this soil, 25 +/- 4 kg ha(-1) yr(-1). Soils, with a texture from sandy loam to clay, showed release rates approaching the net K output, when not K fertilized. According to previous results, the pool of K extractable with 2 M HCl (K-HCl) consists mainly of a fraction of the K in dioctahedral phyllosilicates. In the present investigation K-HCl, corrected for exchangeable K, showed a significant response to soil K balance. More work is needed to yield a precise understanding of the mineralogical nature of this pool and how its measurement might be used to predict potential K release rates in soils. (C) 2007 Elsevier B.V. All rights reserved.