An array of metal oxide sensors has been set up for detecting the vinegary defect in virgin olive oil. The optimization process was carried out evaluating the variables affecting the process by three desirability functions. Repeatability studies for 6 months and within day were done to evaluate the sensor responses and remove those with high relative standard deviation. The sensor responses were preprocessed applying five weight functions previously to build a regression model. Samples of Spanish Arbequina and Picual virgin olive oil varieties spiked with different amounts of acetic acid (15-200 mg/L) were used as a training set for the regression model. The test set was composed of samples of Italian Coratina virgin olive oil spiked with the vinegary standard at five percentages (10, 25, 40, 50, and 75). A fine-adjusted regression coefficient (R(adj)(2) = 0.98) was computed with the test set.