Conventional oil reservoirs are sedimentary rocks that can store and allow the flow of hydrocarbons or water inside. Among the petrophysical properties of reservoir rocks, permeability stands out. During well-logging activities, the Nuclear Magnetic Resonance (NMR) technique is the only well-logging tool that allows estimating the permeability of rocks inside the well. Two NMR permeability models are frequently used by the oil industry: the free fluid model (Timur-Coates or Coates) and the SDR (Schlumberger-Doll-Research). It is observed that there are no dynamic parameters in any of these two most widely used NMR permeability equations. Thus, the objective of this study was to improve the estimation of the permeability of reservoir rocks by Nuclear Magnetic Resonance by adding a dynamic parameter to the SDR model, in order to obtain a new permeability equation by NMR, called K LEAR , where LEAR is the Portuguese acronym of the High Resolution NMR Spectroscopy Laboratory of IFSC/USP. The dynamic parameter chosen was the velocity of magnetization exchange between different sites ( M ), which is observed through NMR exchange experiments. In general, it was possible to observe that there are correlation between the velocity of exchange of magnetization between different sites of the porous media and the permeability, however these correlations are different for sandstones and carbonates.This way, new NMR permeability equations have been proposed, one specific for sandstones and another one for carbonates. These new permeability equations fit better with gas permeability results when compared to the traditional K SDR model. Since the quality and accuracy of information obtained from well-logging has a significant impact on the economic viability assessment of an oil reservoir, the proposed new equations (K LEAR ) have great potential for use by the oil industry, both during well-logging activities as well as laboratory analysis.