SUMMARY Based on new data from permanent and temporary networks, we present fundamental mode Rayleigh wave group velocity maps at periods of 10–150 s related to the lithosphere beneath South America. We analyse waveform data from 1043 earthquakes, from 2002 to 2019, which were recorded by 282 stations. To isolate fundamental mode Rayleigh waves, a phase-matched filter is applied, and the measurements of group velocity are obtained from multiple filter analysis techniques. Thus, we obtain 17 838 paths, covering most of the South American continent, which reach their maximum at the period of 30 s and decrease for both shorter and longer periods. We calculate average dispersion curves and probability density distribution of all measured curves to check the consistency of our data set. Then, regionalized group velocity maps are obtained by iteratively combining the fast marching method and the subspace inversion method. The resolution of our models is assessed by checkerboard tests, which show that the synthetic group velocities are well recovered, despite some amplitude and smearing effects in some portions of the model, probably owing to regularization and uneven ray path coverage. Compared to previous group velocity studies for South America, our models present better resolution, mainly for shorter periods. Our maps of 10 and 20 s, for example, show an excellent correlation with the sedimentary thickness (CRUST1.0) and topography density (UNB$\_$TopoDens). Regions of exposed basement and high-density are related to fast group velocities, while sedimentary basins and low-densities are observed as areas of slow group velocities. We identify small-scale fast group velocity heterogeneities that may be linked to the Rio Apa and Rio Tebicuary cratons as well as to the geochronological provinces of the Amazonian Craton. The most striking feature of our map at 40 s is a fast group velocity structure with the same NE trend of the Transbrasiliano lineament, a Neoproterozoic megashear fault that crosses a large part of the South American continent. Our long-period maps sample lithospheric depths, revealing that cratonic areas of South America, such as the Amazonian and São Francisco cratons, correlate well with fast group velocities. Another interesting feature is the presence of a strong group velocity gradient between the Paraná and Chaco-Paraná basins, which nearly coincides with the location of the Western Paraná Suture, a continental-scale gravity discontinuity. From our group velocity maps, we estimate 1-D S-wave velocity depth profiles at 10 locations in South America: Chaco-Tarija Basin, Borborema Province (BP), Amazonian Craton, Paraná Basin, Tocantins Province, Acre Basin (AcB), Altiplano-Puna Volcanic Complex, Mantiqueira Province (MP), Parnaíba Basin and São Francisco Craton. Most of our inverted S-wave velocity profiles show good agreement with the SL2013sv model at lithospheric depths, except the BP, AcB and MP profiles. Particularly for the BP, a low shear wave velocity, from about 75 to 150 km depth, is a feature that is not present in the SL2013sv model and was probably resolved in our model because of our denser ray path coverage. This decreased S-wave velocity may be due to a lithospheric thinning beneath the BP, as already pointed out by previous studies.