Various approaches for analyzing adsorption data were examined to determine the best method for reporting and interpreting the results of adsorption experiments and ultimately extrapolating laboratory measurements to the field. The interactions of arsenate and goethite were used as representative adsorbate and adsorbent, respectively, although the general principles are applicable to other adsorbate-adsorbent systems as well. A modeling exercise was conducted first to determine the theoretical principles governing the comparison and scaling of adsorption data. These principles were then tested on a suite of experimental data, both new and previously published. Log K-D is significantly more sensitive to variations in adsorbate (As-T) or adsorbent (Fe-T) concentrations than either adsorbed concentration (q) or percentage adsorbed. The sensitivity of K-D relative to q occurs due to the non-linearity of the adsorption isotherm at a given pH, since as the equilibrium aqueous concentration approaches zero, q also approaches zero while K-D approaches infinity. Varying AS(T) and Fe-T while keeping AS(T)/Fe-T fixed yields more consistent values of percentage adsorbed, log K-D, and q, although the adsorbate-to-adsorbent ratios used in laboratory studies often have a rather narrow range compared to those possible in the field. Specific surface area is also a better scaling parameter than the mass of adsorbent, especially between systems with differing adsorbents with markedly different specific surface areas (e.g., natural versus synthetic goethite). Our results have significant implications to contaminant transport modeling, as the constant K-D approach is the most common method of modeling contaminant transport, while contaminant concentrations in the field are typically low, precisely the conditions where K-D is Most sensitive. (C) 2009 Elsevier Inc. All rights reserved.