This study assessed the application of two specialty adsorbents, also known as green sorption media (GSM), including clay–perlite and sand sorption media (CPS) and zero-valent iron and perlite green environmental media (ZIPGEM) to remove long- and short-chain per- and polyfluoroalkyl substances (PFAS) at field scale. The field-scale demonstration employed four GSM filter cells installed near the C-23 Canal (St. Lucie County, FL), which discharges water to the ecologically sensitive St. Lucie River estuary and to the Atlantic Ocean finally. Although prior lab-scale experiments had demonstrated the effectiveness of CPS and ZIPGEM in treating long-chain PFAS, their performance in field-scale application warranted further investigation. The study reveals the critical roles of divalent cations such as Ca2+ and monovalent cations such as ammonium and hydronium ions, as well as other water quality parameters, on PFAS removal efficacy. Ammonia, most likely resulting from photo- and bacterial ammonification, gives rise to elevated ammonium ion formation in the wet season due to the decrease in pH, which ultimately worsens PFAS adsorption. Moreover, there is a strong negative correlation between pH and PFAS removal efficiency in the presence of ammonia, as evidenced by the reduced removal of PFAS during events associated with low pH. Environmental ImplicationsPFAS is one of the forever chemicals. Few studies used recycled material that are cost-effective, adaptable, scalable, and sustainable as specialty adsorptions for PFAS removal. We show the potential value of such filtration media recipes for fit-for-purpose applications using two “green sorption media” mixed with natural and recycled materials for both long-chain and short-chain PFAS removal. Our scientific findings cover the integrative aspect of removal efficiencies and removal mechanisms under changing environmental conditions. It is the first study of its kind to date utilizing a field-scale filtration system to study the PFAS removal mechanisms and convey the corresponding application potential.