The non-destructive assessment and control on and off-line of some quality parameters of the grated cheese is a goal of the dairy industries and distributors. The present research is addressed to explore the possibility of estimating the percentage of moisture, proteins, added rind, and the ripening time of this kind of cheese by a capacitive technique supported by basic statistical analysis and artificial neural networks (ANN). An instrumental chain composed by a parallel plate capacitor probe, for housing the cheese in a suitable container, and a LCR meter, were used to measure the capacitance of different product mixtures, in the frequency range from 100 Hz to 10 kHz. Samples were prepared by mixing cheeses ripened from 8 months to 36 months and rind percentage from 0% to 50%. The results show that moisture and protein percentages can be estimated in a quite good manner with a linear correlation (R-2 up to 0.81) and ANN (R-2 up to 0.83). The ripening time is well linearly assessed in the cheese without the addition of rind (R-2 = 0.994) or if the rind percentage is known (R-2 from 0.872 to 0.963). The rind percentage is very well linearly estimated if the ripening time is known (R-2 from 0.914 to 0.987). Estimation of the rind percentage if the ripening time is unknown appears more difficult (max R-2 = 0.658) because both ripening and addition of rind involve a decrease in capacitance. The frequency at which the capacitance is measured shows only a weak influence in the goodness of the parameters determination. On the whole and as a trend, the correlations improve increasing the frequency. The obtained results, the conformation of the probe and the minimal data treatments mean that this technique can be considered to be applied in the chains of the grated cheese packaging. (C) 2014 Elsevier Ltd. All rights reserved.