Virgin olive oil (VOO) is a source of phenolic compounds, which are related to its recognized biological activity. The analysis of phenolic compounds from VOO begins with the extraction procedure, commonly a liquid-liquid extraction (LLE) with a mixture of methanol:water (60:40, v/v). There are no experiments evaluating if this is the optimal solvent proportion for the maximum phenolics extraction. The aim of this work was to determine the ideal proportion of solvents for the LLE of hydroxytyrosol, tyrosol, oleuropein, (+)-pinoresinol, luteolin and apigenin. Three binary mixture designs were performed. Each experiment was composed by different proportions of the solvents under study (methanol:water; ethanol:water; acetonitrile:water). The contents of the six phenolic compounds, expressed in mg.kg-1 and obtained by HPLC-DAD-FLD, were employed as responses and used for the calculation of mathematical models to predict the optimal extraction condition for each design. All the models were validated through Analysis of Variance (95% confidence level). Simultaneous optimization of the responses was achieved using the desirability function of Derringer and Suich. The optimal conditions predicted by the models were: methanol:water (80:20, v/v), ethanol:water (65:35, v/v) and acetonitrile:water (100:0, v/v). These optimal conditions were validated by means of recovery assays. Recoveries varied from 80.1% (tyrosol) to 102.9% (hydroxytyrosol) (methanol:water design); from 69.0% (tyrosol) to 116.8% (hydroxytyrosol) (ethanol:water design); and from 82.6% (tyrosol) to 112.7% [(+)-pinoresinol] (acetonitrile:water design). The recoveries were statistically compared and it was verified that for luteolin, oleuropein, (+)-pinoresinol and tyrosol, the recovery did not differ significantly, regardless of the optimal mixture. For apigenin, the highest recovery was observed with acetonitrile (100%). The highest recoveries for hydroxytyrosol were achieved with mixtures of methanol:water (80:20, v/v) or ethanol:water (65:35, v/v), which did not differ statistically. Therefore, previous works using methanol:water (60:40, v/v) to extract the phenolics could have underestimated their contents in VOO samples.