The use of zwitterionic surfactants in enhanced oil recovery projects is limited to adsorption on the surface of the reservoir rock, which must be predicted to determine the economic feasibility of these projects. However, there is a lack of models capable of estimating this adsorption and explaining the involved mechanisms. The objective of this study was to provide models that could estimate the adsorption of a zwitterionic surfactant (CAPB) and explain its adsorption mechanisms. Experiments were conducted on carbonate and sandstone rocks using static tests with particulate rock and dynamic tests within rock cores. A methodology was developed to quantify CAPB in brine using high-performance liquid chromatography. As a distinguishing feature, the adsorption was normalized by the specific surface area of the rock, determined through BET analysis (static tests) and microtomography with (micro)CT-scan (dynamic tests). The results were interpreted with empirical and theoretical models integrated with surface potential estimates. For carbonate, it was observed that the first layer of adsorption follows a homogeneous pattern, limited by electrostatic repulsion with the surface, while the second layer follows heterogeneous adsorption, forming surfactant aggregates mediated by hydrophobic interactions between the tails. For sandstone, both layers exhibit a heterogeneous distribution, explaining the higher adsorption between the two rocks. It was concluded that bilayer models are capable of reliably explaining and estimating adsorption under flow conditions, and the surface area was the most relevant factor in the difference of dynamic adsorption between rocks, favored in sandstone.