In this note, we investigate the characteristics of ambient noise cross correlations for station pairs in northern Italy, considering the secondary microseism bandwidth (0.1–0.6 Hz). The preliminary analysis that we performed exploiting the available continuous recording in the investigated area agrees with the recent results of Achauer et al. (2007): the directionality of the noise signal cannot be disregarded when the group velocity is estimated in the range 0.1–0.6 Hz, and the selection of the path orientation for tomography must be carefully performed. In particular, while the favorable directions with respect to microseisms generated along the Atlantic coasts of France, Norway, and the British Islands cover a quite wide azimuthal range (from about 270° N to 5° N), allowing us to reliably estimate the fundamental mode Rayleigh group velocity for paths in the Alps (about 2.7 km/sec), more care must be taken when the microseisms are generated in the Mediterranean Sea. In that case, different locations of the generating areas of microseisms could provide biased estimates of the group velocity due to differences between the true and the apparent velocity of propagation between the stations.