AbstractBench-scale experiments were performed to evaluate poly- and perfluoroalkyl substance (PFAS) desorption from vadose zone soils collected at a site historically impacted with aqueous film-forming foam (AFFF). A sequential dilution batch method was employed to obtain PFAS desorption isotherms from a shallow surface soil [0.03 to 0.9 m below ground surface (bgs)] and from a deep soil (0.9–2.4 m bgs). Results showed that the desorption isotherms were reasonably described by a linear model, and that a fraction of the soil-bound PFAS mass was not readily desorbed (or, non-labile); the recalcitrance of this non-labile fraction was evaluated on a subset of samples and shown to persist for up to 9 months. The non-labile fraction of PFAS mass sorbed to the soil (non-labile PFAS sorbed mass divided by the total PFAS sorbed mass) ranged from 0.17 g g−1 for perfluorohexane sulfonamide in the deep soil to 0.87 g g−1 for 8∶2 fluorotelomer sulfonate in the shallow soil, and generally increased with the PFAS organic carbon:water partitioning coefficient (Koc). The fraction of the PFAS within the non-labile fraction was generally greater in the shallow soil than in the deep soil, likely due to the elevated organic carbon content of the shallow soil. The slopes of the desorption isotherms associated with the readily-desorbable (labile) PFAS fraction for the shallow soil were within the ranges predicted by Koc-based models derived from adsorption data. Overall, results for the soils studied herein suggest that much of the PFAS mass present in the soil likely is not readily available for leaching into infiltrating waters. Such information could provide useful insight for developing PFAS soil cleanup standards.