Earthquakes have been studied by means of seismometers recording the elastic waves travelling through the interior of our planet. Global Navigation Satellite System and Synthetic Aperture Radar surveys, measuring surface displacements, have provided additional information on earthquakes, as well as on those solid Earth processes responsible for them, such as subduction, collision and extension and the inter-seismic strain accumulation. This instrumentation is deployed over land and thus misses the seas, often surrounding regions where large earthquakes occur. This limitation is nowadays overcome by space gravity missions, thanks to their uniform coverage of the Earth, both inland and offshore. In this perspective, Gravitational Seismology has been identified as a new application of the Next-Generation Gravity Mission (NGGM), with the aim of evaluating its overall performance and of assessing the detectability of earthquake gravity signatures, as well as of those from active tectonics and inter-seismic deformation. Within the framework of self-gravitating viscoelastic Earth models, we have simulated the co- and post-seismic gravity signatures of 291 scenario earthquakes, with different occurrence times and geographical locations, focal mechanisms, depths and lines of strike, and included into the background gravity feeding the NGGM closed-loop simulation which provides observables of multiple pairs of GRACE-like satellites, given the instrument noise. NGGM earthquake detectability is herein defined on the possibility of estimating the amplitude of the original gravity signature of each earthquake by inversion of synthetic NGGM gravity data, consisting of 156 28-day gravity field solutions (about 11 years). For about two thirds of earthquakes of magnitude as low as 7, comparable with the 1980 Irpinia intraplate earthquake, the amplitudes have been estimated with a relative error less than 10% (and less than 50% for almost all the earthquakes), assuming as known the time variable contributions from atmosphere, oceans, hydrology, continental ice and glacial isostatic adjustment. When these contributions are inverted simultaneously with the earthquake ones, instead, we have had to increase the earthquake magnitude to 7.8 in order to estimate more than half of their amplitudes with a relative error less than 10%. We thus have shown that the NGGM will be able to detect, in most cases, the co- and post-seismic signatures of earthquakes of at least magnitude 7.8 and that this lower magnitude threshold can decrease down to magnitude 7 by improving the modelling of the background gravity field.