We analysed short-period seismograms from about 250 very local events recorded by a network of high dynamic range short-period seismic stations deployed on Mt Vesuvius to estimate the site-corrected short-period seismic attenuation. We calculated Q(C)(-1), the inverse of the quality factor for coda waves, for short lapse times (12 s), and Q(P)(-1) and Q(S)(-1), the inverse of the direct body wave quality factors for P and S waves following shallow rah paths. We used the single scattering assumption to fit the amplitude envelopes of the coda at different frequency bands, obtaining a Q(C)(-1) slowly varying with frequency, similar to values measured for other volcanoes at the same lapse time. Site-corrected Q(P)(-1) and Q(S)(-1) were estimated using the frequency decay method for both P and S waves in two frequency bands, 1-6 Hz and 15-24 Hz. Results show that Q(P)(-1) is 0.028 in bot frequency bands, and Q(S)(-1) is 0.015 and 0.017 at low and high frequencies respectively, almost independent of the frequency, and that the value of Q(P)(-1) is about twice that of Q(S)(-1), as measured worldwide. A check was made by measuring the broadening of the first P-wave pulse with station-source distance, finding Q(P)(-1) = 0.024. Using the independent estimates of Q(C)(-1) and Q(S)(-1), we separated the intrinsic from the scattering quality factor at Mt Vesuvius, obtaining a high scattering and a low intrinsic Q(-1), as already observed for other volcanoes. This result confirms the hypothesis of the predominant role of the scattering phenomena with respect to the intrinsic dissipation in the seismogram formation for volcanic earthquakes.