Mass composition analysis is a fundamental task to test any theoretical model concerning the origin and the nature of the primary cosmic ray radiation at the highest energies. Different energy spectra are predicted to be observed at ground by the present theories, according to the mass of the primary particle, so the knowledge of the energy spectra for every mass component, or at least for groups of components, is required in order to discriminate among the proposed models. At lower energies (E< 1014 eV) the composition of cosmic rays can be measured using direct detection techniques, such as spectrometers and calorimeters. At higher energies, the measurement of the mass is generally performed by indirect techniques, which make use of parameters sensitive to the primary mass, and determined by the shower development in the atmosphere. Among such parameters, Xmax (the depth at which the longitudinal shower has its maximum), Nmax (the number of shower particles at Xmax) and Nµ (the number of muons at a given distance from the shower axis) are widely used. In the knee region (1015÷ 1017 eV) a recent analysis from KASCADE experiment, based on the deconvolution of a 5-component mass spectra starting from the experimental Ne-Nµ scatter plots, shows that the knee is due to a decrease of the light component with respect to the heavier one, and that the knee position for higher masses shifts towards higher energy [1]. A clear increase of the mean logarithmic mass as a function of the primary energy is found in other experiments, such as EASTOP-MACRO [2]. While the experimental results show a definite trend in this …