A previous study showed that large differences in K uptake by alfalfa (Medicago sativa L.) grown in the greenhouse occurred on soils which had similar ammonium-acetate-extractable K (NH4OAc-K) concentrations at seeding. Differences in total K uptake appeared to be closely related to the uptake of nonexchangeable K (NEK). The objective of this study was to determine what fractions or forms of NEK in the soils were responsible for the large differences in total K uptake noted in the greenhouse experiment. The total amount of NEK extracted in seven sequential extractions with boiling 1 M HNO3 (Mactotal) varied widely among soils and was significantly related to the total amount of K (r = 0.69; P < 0.05) and NEK (r = 0.75; P < 0.05) taken up by eight crops of alfalfa. Mactotal was separated into two fractions of differing solubility and the most soluble fraction, StepK, was highly correlated with NEK uptake (r = 0.85; P < 0.01) and total K uptake (r = 0.86; P < 0.01). The least soluble fraction, CRK, was not significantly correlated with either NEK or total K uptake. A multiple regression equation with NH4OAc-K and StepK as independent variables explained 86% of the variation in total K uptake among soils. The coarse clay fraction (0.2–2.0 μm) was the main source of NEK taken up by alfalfa. It was concluded that differences in the K-supplying capacity of southern Ontario soils were due not only to differences in NH4OAc-K concentrations but also to the amount of the StepK fraction of NEK present in the soils. The quantity of StepK present appeared, from a field experiment, to be influenced by the amount of fertilizer applied over a period of years. If a simple method of measuring StepK can be found, soil tests for plant available K in Ontario soils might be greatly improved. Key words: HNO3 extractable K, nonexchangeable K, NH4OAc-K, alfalfa, coarse clay, fine clay