The electronic tongue multisensor system was applied to the measurements in about 160 samples of the Port wines of different ages (from 2 to 70 years old) and of different types including Tawnies, Vintages, Late Bottled Vintage (LBV) and harvest wines. Multivariate calibration for the prediction of the Port wine age was made applying PLS regression to the electronic tongue and chemical analysis data. Orthogonal signal correction (OSC) was evaluated as data pre-processing method. OSC filtering was applied to the both data sets and the results were compared. The age of the Port wine was predicted with similar accuracy using both electronic tongue and chemical analysis data. The accuracy of age prediction was improved to about 1.5 years when using only the samples aged from 10 to 35 years for the calibration. This appears to be acceptable for the practical applications.