Combined analysis of seismic P and S velocity information provides a reasonable and efficient basis for the petrologic interpretation of seismic cross sections. In this paper, a methodology is presented which allows extraction of prominent features related to well-defined P velocities and Poisson's ratios from a tomographic velocity model using a classification approach. We used first-arrival travel time data from a near-vertical seismic experiment and independently determined P and S velocities by forward and inverse modeling. Resolution and uncertainties were estimated from inverting synthetic data. The classification procedure was carried out in two subsequent steps. First, prominent classes were identified in the parameter space spanned by Poisson's ratios and P wave velocities. For this purpose, a probability density function was calculated from the tomograms. A function measuring the topography of the probability density was then determined, and a histogram analysis was carried out to detect significant classes. In the second step, the results from principal component analysis for the identified classes were used to map their distribution along the seismic profile. We applied the method to the Messum intrusive complex of Namibia and identified three prominent classes. On the basis of the integration of petrophysical data and comparison with surface geology, we conclude that quartz-syenite composition dominates the upper 800 m of the crust under the complex. Outside of the intrusion the upper crust has properties corresponding to felsic metasediments and granites which are abundant in the local basement. This material shows strong depth-dependent changes of seismic properties which are ascribed to decreasing porosity and fluid saturation with depth.