Elsewhere we have shown the possibility to follow post harvest evolution of strawberry by Proton Transfer Reaction-Mass Spectrometry (PTR-MS) and we indicated how this technique can produce a rapid and non-destructive fingerprint of the head-space of different agroindustrial products (Boschetti et al., 1999, Biasioli et al. 2003a). In particular we recently showed that the coupling of PTR-MS with proper data mining techniques unambiguously identifies the variety of single intact strawberry fruits (Biasioli et al. 2003b). In this latter work we considered only a few samples for 9 varieties harvested in 2002. Here we show that i) different data mining techniques can be effectively applied to PTR-MS data and ii) variety discrimination is evident even if we increase the number of fruits and extend the sampling on two different years. Data refer to two commercial strawberry cultivars: 'Miss' and its daughter 'Queen'. Two supervised chemometric techniques (Discriminant Analysis - both Linear and Quadratic - and Artificial Neural Networks) have been used to authenticate the cultivar of 63 strawberry fruits, collected in different places and at different times, on the basis of their PTRMS spectra. The optimized models, built using only the 2 most discriminating variables, have been able to correctly predict 100% of the samples as evaluated by a leave-n-out cross-validation procedure, with n ranging from 1 to 6.