The cosolvency model was not applicable for predicting the sorption of organic carboxylic acids. The reason of inapplicability was investigated by analyzing the solubility (Sm) and sorption (Km) of benzoic acid, 2,4-dichlorophenoxyacetic acid (2,4-D), and 2,4,6-trichlorophenol (2,4,6-TCP). The Sm and Km by two iron-rich soils was measured as a function of methanol volume fraction (fc), electrolyte compositions, and pH(app). For 2,4,6-TCP, the Km of both neutral and anion species was well-explainable by the cosolvency model, exemplifying the knowledge of cosolvency power (σ) being sufficient to describe its sorption. However, for benzoic acid and 2,4-D, the Km of organic anions increased with fc, illustrating the organic carboxylate to be responsible for the deviation. The Sm of organic anions was not affected by the ionic valence (Ca(2+) vs. K(+)) of liquid phase. Among hydrophilic quantities of the 2,4-D sorption, the fraction of anion exchange increased with fc while the fraction of Ca-bridge decreased in the same range. Adding solvent in soil-water system is likely to render soil surface charge more positive, fortifying the anion exchange, but opposing the formation of Ca-bridging. Therefore, it can be concluded that the positive Km-fc relationship is due to the anion exchange of organic carboxylate with positively charged soil surface, whose contribution is >50% of overall sorption at solvent-free system and becomes greater with fc up to 82%.