We present a synthesis and a comparison of the results of two temporary passive seismic experiments installed for a few months across Central and Northern Zagros. The receiver function analysis of teleseismic earthquake records gives a high-resolution image of the Moho beneath the seismic transects. On both cross-sections, the crust has an average thickness of 43±2 km beneath the Zagros fold-and-thrust belt and the Central domain. The crust is thicker beneath the hanging wall of the Main Zagros Reverse Fault (MZRF), with a larger maximum Moho depth in Central (69±2 km) than in Northern Zagros (56±2 km). The thickening affects a narrower region (170 km) beneath the Sanandaj-Sirjan zone of Central Zagros and a wider region (320 km) in Northern Zagros. We propose that this thickening is related to overthrusting of the crust of the Arabian margin by the crust of Central Iran along the MZRF considered as a major thrust fault cross-cutting the whole crust. The fault is imaged as a lowvelocity layer in the receiver function data of the Northern Zagros profile. Moreover, the crustal-scale thrust model reconciles the imaged seismic Moho with the Bouguer anomaly data measured on the Central Zagros transect. At upper mantle depth, P-wave tomography confirms the previously observed strong contrast between the faster velocities of the Arabian margin and the lower velocities of the Iranian micro-blocks. Our higher-resolution tomography combined with surface-wave analysis locates the suture in the shallow mantle of the Sanandaj-Sirjan zone beneath Central Zagros. The Arabian upper-mantle has shield-like shear-wave velocities, while the lower velocities of the Iranian upper mantle are likely due to higher temperature. But these velocities are not low enough and the low-velocity layer not thick enough to conclude on a delamination of the lithospheric mantle lid beneath Iran. The lack of a high-velocity anomaly in the mantle beneath Central Iran suggests that the Neotethyan oceanic lithosphere is probably detached from the Arabian margin.