The adsorption of binary mixtures containing particles A and B on homogeneous substrates is studied by Monte Carlo (MC) simulations, quasi-chemical approximation (QCA), and exact counting of states on finite cells (we call this approach cluster approximation, CA). The energies involved in the adsorption model are five: (1) $$\epsilon_A,$$ ϵ A , interaction energy between an A particle and a lattice site; (2) $$\epsilon_B,$$ ϵ B , interaction energy between a B particle and a lattice site; (3) $$w_{AA},$$ w A A , nearest-neighbor interaction energy between two A particles; (4) $$w_{AB}$$ w A B (= $$w_{BA}$$ w B A ), nearest-neighbor interaction energy between an A particle and a B particle and (5) $$w_{BB}$$ w B B , nearest-neighbor interaction energy between two B particles. The process is monitored by following the coverage of both species with the simultaneous increasing of the individual chemical potentials of each mixture component. A non-trivial interdependence between the partial adsorption isotherms was observed and discussed in the context of the lattice-gas theory. The theoretical formalism is used to model experimental data of methane-carbon dioxide mixtures adsorbed on activated carbon. In addition, an excellent agreement was obtained between theoretical and MC simulation results. This finding evidences the usefulness of CA and QCA as a starting point to predict the behavior of a system governed by a large number of parameters.