SUMMARYPotassium release from saturated spherical aggregates of three clay soils to Ca‐resin was measured and compared with release in suspensions (Talibudeenet al., 1978). The diffusion coefficient for K was calculated from measurements of the tortuosity factor using36Cl, volumetric water content and buffer power. Release of K is in the order: suspensions>2–3 mm>6–8 mm aggregates. A computer model incorporating diffusion analysis and release of K shows that in the aggregates the rapidly diffusing exchangeable K, Kex, is best measured by the extrapolated intercept of a desorption isotherm (24 h equilibration), and is 8–55% less than that extracted by NH4OAc leaching. Release of K within the aggregates (Knex, a fraction approximating to Talibudeen's peripheral K) is delayed until after about 95% of Kexhas been depleted which takes 1–2 days. The effective release rate of Knexinside the aggregates is ∼ 17% less than in soil suspensions. Diffusion imposes only small limitations on the release from aggregates of Knex. This rate of release depends on the frequency of re