The Cenozoic Baikal Rift Zone (BRZ) is situated in south-central Siberia in the suture between the Precambrian Siberian Platform and the Amurian plate. This more than 2000-km long rift zone is composed of several individual basement depressions and half-grabens with the deep Lake Baikal at its centre. The BEST (Baikal Explosion Seismic Transect) project acquired a 360-km long, deep seismic, refraction/wide-angle reflection profile in 2002 across southern Lake Baikal. The data from this project is used for identification of large-scale crustal structures and modelling of the seismic velocities of the crust and uppermost mantle. Previous interpretation and velocity modelling of P-wave arrivals in the BEST data has revealed a multi layered crust with smooth variation in Moho depth between the Siberian Platform (41 km) and the Sayan-Baikal fold belt (46 km). The lower crust exhibits normal seismic velocities around the rift structure, except for beneath the rift axis where a distinct 50-80-km wide high-velocity anomaly (7.4-7.6 ± 0.2 km/s) is observed. Reverberant or "ringing" reflections with strong amplitude and low frequency originate from this zone, whereas the lower crust is non-reflective outside the rift zone. Synthetic full-waveform reflectivity modelling of the high-velocity anomaly suggests the presence of a layered sequence with a typical layer thickness of 300-500 m coinciding with the velocity anomaly. The P-wave velocity of the individual layers is modelled to range between 7.4 km/s and 7.9 km/s. We interpret this feature as resulting from mafic to ultra-mafic intrusions in the form of sills. Petrological interpretation of the velocity values suggests that the intrusions are sorted by fractional crystallization into plagioclase-rich low-velocity layers and pyroxene- and olivine-rich high-velocity layers. The mafic intrusions were probably intruded into the ductile lower crust during the main rift phase in the Late Pliocene. As such, the intrusive material has thickened the lower crust during rifting, which may explain the lack of Moho uplift across southern BRZ.