Batch tests and different statistical tools of data analysis were used to re-evaluate the overall effect of soil characteristics and liquid phase composition on the extent of phenanthrene adsorption in complex soil-water systems. The linear isotherm models was capable of adequately describing the equilibrium data under extremely varying conditions of soil type, environmental conditions (pH, temperature, ionic strength) and amendments (surfactant, oil, dispersing agent, glucose). Consistent with existing mechanistic models, the multivariate approach also identified the organic carbon content (f(oc)) of soil as the key parameter controlling the phenanthrene adsorption constant (K-d) in nonamended systems (K-oc was 17,700 mL/g). From studying the effect of the amendments, two interactions (surfactant-pH and surfactant-oil) and two main effects (surfactant and oil) have been detected. An empirical linear model of K-d as a function of f(oc) pH, oil content of soil, and surfactant dose was developed for the range of conditions studied. The proposed model and modeling approach can be adapted to other types of contaminants or variables for specific natural and engineered systems.