All predictive toxicological models concern individual chemicals. An exposure to a mixture of chemicals is the reality, however. Toxicity of an individual chemical itself or in a mixture with other chemicals may significantly differ. The presentation shows our attempt to find a suitable model for a prediction of acute toxicity of binary mixtures of organic chemicals. We have suggested a molar ratio as a descriptor of the mixture composition and R-plot for a graphical representation of the relationship between acute toxicity and the mixture composition (QCAR - Quantitative Composition - Activity Relationship). The approach was inspired by Rault and Dalton laws, their positive and negative deviations in a mixture behavior of real gases, Loewe and Muschnek isoboles and Finney test of additivity. The inhibition of movement of worms Tubifex tubifex has been measured as the acute toxicity with the mixtures benzene - aniline, benzene - ethanol and benzene - nitrobenzene as representatives of the three main types of interactions: potentiation, inhibition, addition. As a physicochemical partner, distribution of components of the mixture between its gaseous and liquid phases has been chosen.. All the relationships should be express in a mathematical function to allow a prediction knowing a composition of a mixture and physicochemical properties of the components.