We report on a passive seismic experiment on a historical tower in northern Italy. Assessment of dynamical properties of such structures is known to be very important to anticipate their response to possible earthquake excitation. We show how classical seismological analyses can image, in high definition and detail, many modes of vibration, continuously excited by background tremor due to vehicle traffic. Similarly to what Earth’s normal modes reveal, such information pose constraints on the elastic (and anelastic) structure of a building that, in the case of a historical edifice, are not generally known, because appropriate information on its fabrication and constitution is not available. Our case refers to the twelfth century, 97 m tall, Asinelli tower in Bologna (northern Italy)—the tallest slender masonry building in Europe. We detect and analyze several modes of free flexural vibration, besides compressional and torsional modes. Free vibrations occur with slightly different natural frequencies along two orthogonal directions, as a consequence of a discrepancy between centers of mass and stiffness. This is apparent by the splitting of frequencies of modes of vibration. For each mode, the polarization of particle motion shows that energy cyclically transfers between the two different degrees of freedom of the system. The spectral signature of the nearby Garisenda tower can also be recognized on the spectrum of Asinelli, and vice versa, but the frequencies differ, so there is no cross influence between the two. These data can be used to calibrate the parameters used for dynamic modeling, to investigate variations in time of the dynamic response, and to characterize the anthropogenic sources of ground motion to possibly mitigate their effects. Although the engineering community has extensively addressed the field of structural health monitoring with specific tools, we show that methods part of any seismologist toolbox can also provide detailed information about buildings that can be used for their numerical modeling.