In this thesis a physics-based constitutive theory for sedimentary porous rocks is proposed combining the results of laboratory tests, theoretical analysis, and numerical validation. The motivation for this framework stems from triaxial experiments on calcarenite performed up to 50% axial strain inside an X-Ray CT-scan. These tests revealed that: 1) calcarenite plastified at the first increment of displacement; 2) with increasing axial strain, the material underwent a phase change where all the inter-granular bonds broke, the pore space collapsed and the material behaved as sand; 3) in repeating loading-unloading cycles and relaxation tests, the deformation of the rock became increasingly more rate-dependent with strain, as a result of the aforementioned phase change and reorganization of released grains. Motivated by these experiments, a visco-plastic flow law is proposed. The viscosity of the material is assumed to be a function of the temperature, pore-pressure and energy required to alter the inter-granular interfaces. Thus, stress equilibrium and flow law are fully coupled to the energy and mass conservation laws, constituting a closed system of equations. In order to solve this system, the theoretical framework is implemented into the tightly coupled Finite Element code REDBACK, and its qualitative behaviour is analysed in monotonous and cyclic isotropic compression as well as in direct shear for different loading rates. A series of numerical calibration tests against different types of rocks (sandstone, mudstone, calcarenite), saturating conditions (dry, wet) and stress paths (triaxial, isotropic) is then performed, concluding that the mechanical response of sedimentary porous rocks in strains usually achieved in laboratory testing is determined by the strength of the cementitious material bonding the grains. The latter is shown to be stress path dependent under the hypotheses made in this thesis and the interfaces are shown to obey a Kelvin-like law at the microscopic level. Finally, the proposed framework is applied at geophysical scale problem and is qualitatively linked with theoretical studies of landslide and faults in the literature. A reinterpretation of the brittle to ductile transition is then attempted linking the two cases (brittle and ductile) to the types of instabilities that the model theoretically predicts.