Ambient vibrations were recorded by two dense short-period seismic arrays located on Mount Vesuvius. The aim of this study is to derive the surface velocity structure through the application of array techniques and to interpret it in terms of resonance effects. Mount Vesuvius, which is located in proximity to the Apennines earthquake source-zone, is one of the most dangerous volcanoes of the world. We compared different array techniques to derive the dispersive properties of the surface waves composing the noise wave field. The frequency-wavenumber (f-k) spectral method applied to the data recorded by array A furnished the Rayleigh waves and Love waves dispersive functions and proved the time–space stationarity of the noise wave field. The existence of a stochastic wave field justifies the use of the spatial autocorrelation (SPAC) method and the time stationarity autocorrelation (TSAC) method, which appear to be most effective at the lowest frequencies. The TSAC method, which time averages, is confirmed as the most user-friendly, because it does not require circular geometries. The knowledge of the surface structure is an important goal in site-effects studies and is used to determine the frequency response of the near-surface geology. Velocity models have been derived from the dispersion curves, using both trial and standard inversion procedure. The results are compared with those obtained from array measurements in adjacent areas (Scarpa et al., 2003), revealing similar mechanical properties in the cover deposits and heterogeneities down to 40 m of depth. The shear-wave velocity profiles derived at the array-sites are used to model the 1D transfer functions for vertically incident shear waves. The fundamental resonance frequencies agree with the peak frequencies observed in the microtremor horizontal-to-vertical spectral ratios. The frequency band of amplification is also compatible with the results obtained from local earthquakes at sites located at the same elevation on Mount Vesuvius (Galluzzo et al., 2009).