AbstractAdsorption of perfluoroalkyl acids (PFAAs) by granular activated carbon (GAC) was evaluated in bench‐, pilot‐, and full‐scale studies to determine effects of PFAA characteristics and background organic matter on carbon use rates. Rapid small‐scale column tests (RSSCTs) were conducted according to the proportional diffusivity (PD) design to assess their suitability to predict full‐ or pilot‐scale GAC performance. PFAA removal from groundwater (GW) and coagulated surface water (SW) was studied with two sub‐bituminous coal‐based GACs. In batch tests conducted with pulverized GAC, the GACs performed similarly in GW, but the GAC with the larger mesopore volume was more effective for PFAA removal from SW. In column tests, carbon use rates decreased with increasing PFAA chain length and were lower for GW (total organic carbon [TOC] = 0.7 mg/L) than for SW (TOC = 2.0–2.7 mg/L). The volume of SW that could be treated to 10% or 50% PFAA breakthrough was about 50–60% of the volume of GW that could be treated when comparing pilot‐scale data for SW with full‐scale data for GW. Consistent differences in PFAA adsorption capacity were not observed for empty bed contact times of 13 and 26 min in full‐scale adsorbers treating GW. The PD‐RSSCT simulating PFAA removal from GW consistently overpredicted full‐scale adsorption capacity, on average by ~70%. Using a carbon use rate of <25 mgGAC/Lwater treated) as a criterion for the feasibility of GAC treatment, full‐ and pilot‐scale GAC adsorber data suggest that GAC is a viable treatment option (carbon use rate < 25 mgGAC/Lwater treated) for perfluoroalkylcarboxylic acids with six or more carbon atoms in SW and five or more carbon atoms in GW. For perfluoroalkyl sulfonic acids, GAC treatment is viable for compounds containing four or more carbons based on results obtained with both SW and GW.