The Mohorovicic discontinuity is the boundary between the Earth’s crust and mantle. Several isostatic hypotheses exist for estimating the crustal thickness and density variation of the Earth’s crust from gravity anomalies. The goal of this article is to compare the Airy-Heiskanen and Vening Meinesz-Moritz (VMM) gravimetric models for determining Moho depth, with the seismic Moho (CRUST2.0 or SM) model. Numerical comparisons are performed globally as well as for some geophysically interesting areas, such as Fennoscandia, Persia, Tibet, Canada and Chile. These areas are most complicated areas in view of rough topography (Tibet, Persia and Peru and Chile), post-glacial rebound (Fennoscandia and Canada) and tectonic activities (Persia). The mean Moho depth provided by CRUST2.0 is 22.9 ± 0.1 km. Using a constant Moho density contrast of 0.6 g/cm3, the corresponding mean values for Airy-Heiskanen and VVM isostatic models become 25.0 ± 0.04 km and 21.6 ± 0.08 km, respectively. By assuming density contrasts of 0.5 g/cm2 and 0.35 g/cm3 for continental and oceanic regions, respectively, the VMM model yields the mean Moho depth 22.6 ± 0.1 km. For this model the global rms difference to CRUST2.0 is 7.2 km, while the corresponding difference between Airy-Heiskanen model and CRUST2.0 is 11 km. Also for regional studies, Moho depths were estimated by selecting different density contrasts. Therefore, one conclusion from the study is that the global compensation by the VMM method significantly improves the agreement with the CRUST2.0 vs. the local compensation model of Airy-Heiskanen. Also, the last model cannot be correct in regions with ocean depth larger than 9 km (e.g., outside Chile), as it may yield negative Moho depths. This problem does not occur with the VMM model. A second conclusion is that a realistic variation of density contrast between continental and oceanic areas yields a better fit of the VMM model to CRUST2.0. The study suggests that the VMM model can primarily be used to densify the CRUST2.0 Moho model in many regions based on separate data by taking advantage of dense gravity data. Finally we have found also that the gravimetric terrain correction affects the determination of the Moho depth by less than 2 km in mean values for test regions, approximately. Hence, for most practical applications of the VMM model the simple Bouguer gravity anomaly is sufficient.