<p>This study proposes a novel method for estimating the surface tension (ST) of binary and ternary liquid mixtures based on the Eyring’s theory. Initially, four weighted average models based on mole and mass fractions were evaluated using ideal solution systems. The results confirmed that the Eyring-based weighted average models successfully reproduces experimental ST values. For non-ideal systems, we developed the “Wilson-SurTen model” by incorporating the Wilson model as an excess Gibbs energy model to describe excess ST based on the Eyring’s theory. Model parameters were regressed using the experimental ST data of various binary systems, including both non-aqueous and aqueous systems. The proposed model demonstrated superior correlation performance, particularly in aqueous systems where conventional models often fail. Furthermore, STs for ternary systems were predicted only using binary parameters determined from the constituent binary systems, without additional ternary interaction parameters. The results indicate that the Wilson-SurTen model significantly improves estimation accuracy, particularly for non-ideal aqueous systems. Furthermore, the mass fraction-based approach consistently outperformed the mole fraction-based one in terms of predictive capability.</p>