Release of non-exchangeable potassium (NEK) in soil is closely related to bioavailability of the potassium. The aim of this study was to characterize release and bioavailability of NEK in 8 soils, using pot K depletion experiments, 0.2 mol L^-1sodium tetraphenylboron(NaTPB) extraction method and kinetic models. From each soil, 3 portions, 0.5 g each as replicate, were taken out for incubation in 3 ml of extracting solution (0.01 mol L^-1 EDTA and 0.2 mol L^-1 NaTPB) separately for a period varying from 0.5 to 96h. Four kinetic models (First-order, Elovich, power function and parabolic diffusion) were used to describe NEK release characteristics. Results show that NEK release rate in soil varied sharply with the agro-ecological region, being the highest in Northwest and Northeast China, and the lowest in Central and East China. NEK release in soil could be divided into two phases, fast and slow ones. The fast phase occurred at the beginning of the release, when NEK at the edges and wedge zones of clay mineral interlattices broke up bondage of the minerals by diffusion, while the slow phase came late when NEK inside the interlattices became the major source of K release. All the four models, the first-order (R2 = 0.817~0.926, SE = 11.9~215.1), Elovich equations (R2 = 0.952~0.997, SE = 4.6~66.9), power function (R2 = 0.869~0.990, SE = 3.9~127.8) and parabolic diffusion (R2 = 0.790~0.963, SE = 7.5~211.4), were good enough to describe NEK release kinetics. However, in choosing a proper one, it is essential not only to compare them in fitting degree, but also take into account their practicality and physical significance of the information they may provide. The Elovich equation failed to describe the early period of NEK release adequately, while the parabolic diffusion model yielded more than one NEK release rates. Thus, the NEK release rate described by the power function model might be a good indicator of NEK potential of soil. By such an indicator, the 8 soils were found to be declining in NPK release potential from west to east and from north to south of the country. Ryegrass were grown for 8 harvests in a pot experiment to evaluate K supplying capacity of the 8 soils. It was found that soil K supplyingpotential consisted with soil K release capacity, and that in soils high in K release amount and rate, the ryegrass without any K fertilizer applied was still quite high in relative and cumulative biomasses, cumulative K uptake and relative K content; and vice versa. The soil NEK release rates described with the four kinetic models was significantly related to K content and K uptake of the ryegrass without K fertilizer applied. In order to judge whether K supply of a soil was adequate or not, relationships between relative biomass and K content of the ryegrass without K fertilizer applied and between K content of the ryegrass and soil NEK release rate were determined. Based on the relationships, 90% of the relative biomass of the ryegrass without K fertilizer appliedwas set as threshold of the index of soil K nutrition, equaling to 34.0 g kg-1in soil K content and 396 mg kg-1 h-1 in soil NEK release rate. It is, therefore, concluded that out of the 8 soils, only Aridosol in Xinjiang and Isohumosol in Shaanxi and Heilongjiang can ensure ryegrass a good yield without showing any potassium deficiency symptomin a short termwithout K fertilizer application.