An optimisation protocol is presented for the resolution of complex mixtures in isocratic RPLC with binary mobile phases of organic solvent and water, which is based on the prediction of peak position and shape of the individual compounds. A good description of the retention was achieved through the application of statistical weights to the widely used linear or quadratic relationships between the logarithm of the retention factor (log k) and the organic solvent concentration in the mobile phase. The maximisation of the product of peak purities for each compound is shown as a competitive resolution strategy versus the worst value of a selectivity parameter. Peak purities allow one to associate a single resolution value to each compound, which is not affected by the identity of the interfering peaks. It is shown how when full resolution is not achieved with a single mobile phase, the same experimental data set (retention factors, asymmetries and efficiencies) can be used for finding two or three optimal complementary mobile phases (CMPs). Each CMP resolves fully some compounds in the mixture, while the remaining compounds can overlap among them. The elementary limiting resolutions, which measure the maximal separation degree for each compound, are also given as a useful guide in the selection of the elution conditions. A mixture of 13 phenols (phenol, chloro-, bromo-, nitro- and methyl-derivatives), eluted with acetonitrile-water or methanol-water mobile phases, is used to show the proposed methodology.