In this thesis I present a geophysical and petrological study that aims to define the lithosphere structure and the chemical variations in the compositions in the lithosphere mantle in the Iberian plate, the western Mediterranean and North Africa. Also, the low Pn velocities measured in the western Mediterranean basins and the sub-lithospheric low-velocity anomaly imaged from tomography models in the Iberian plate and North Africa and the relationship with high mean topography are investigated. I present the crust and upper mantle structure along four transects, crossing the NE-Iberian Peninsula, the Western Mediterranean basin and the Algeria margin and ends at the Tell–Atlas Mountains, two profiles crossing the entire Iberian plate in N-S direction and another transect across the southern Central Iberian Zone, the Gibraltar Arc System and the Atlas Mountais. The modeling in the present work is based on an integrated geophysical-petrological methodology combining elevation, gravity, geoid, surface heat flow, seismic and geochemical data. Unlike previous models proposed for the region where the density of the lithospheric mantle is only temperature-dependent, the applied methodology allows inferring seismic velocities and density in the mantle down to 400 km depth from its chemical composition through self-consistent thermodynamic calculations. The crustal structure is mainly constrained by active-source and passive-source seismic experiments, whereas the upper mantle is constrained by tomography models. In lithospheric mantle composition, this methodology allows us to reproduce the main trends of the geophysical observables as well as the inferred P-, Pn- and S-wave seismic velocities from tomography models and seismic experiments available for the study region. The first geo-transect is a new model on the present-day lithospheric structure along a 1100 km crossing the NE-Iberian Peninsula, the Western Mediterranean basin and the Algeria margin and ending at the Tell–Atlas Mountains. Five lithospheric mantle compositions are considered including predominantly average Phanerozoic and lherzolitic Proterozoic in the continental mainland, and more fertile PUM (primitive upper mantle) compositions in the Western Mediterranean basin. The second profile running from the North-Iberian Margin to the Duero basin and is extended to the Alboran basin in the third profile. In this case I have considered four lithospheric mantle compositions: a predominantly average Phanerozoic in the continental mainland, two more fertile compositions in the Alboran Sea and in the Calatrava Volcanic Province, and a hydrated uppermost mantle in the North-Iberian Margin. In the fourth geo-transect, I defined three distinct chemical lithospheric mantle domains: a predominantly average Phanerozoic composition in the continental mainland, and two rather fertile mantle compositions beneath the Volcanic areas, close to the Calatrava Volcanic Province and the Atlas Mountains. Mantle petrology affects the resulting density distribution and LAB (lithosphere–asthenosphere boundary) geometry and allows a direct comparison with tomography models and seismic data. The obtained LAB geometries in lithosphere show large variations in lithospheric mantle thickness. Deepest LAB geometries are found below the Pyreenes, Cantabrian Mountains, Betics Range and Strait of Gibraltar while the shallowest LAB is reached in the Mediterranean basins (Valencia Through, Balearic Promontory, Algerian and Alboran basins). The LAB depth beneath the Iberia Meseta is deeper than in previous studies while in the Atlas Mountains the resulting LAB is relatively shallow as expect from others studies. Measured low Pn velocities in the Western Mediterranean basin can be explained by either serpentinization and/or seismic anisotropy and only partly by transient thermal effects whereas in the Bay of Biscay is explained by only by serpentinization. Also, the sub-lithospheric negative anomaly imaged in tomography models below the Iberian plate and the Atlas Mountains, is related to the high mean topography in these areas and can be partly consistent with a low-velocity/high-temperature/low-density modelled layer in the sublithospheric mantle.