Longitudinal and transverse nuclear magne c resonance (NMR) relaxa on signatures in porous rock were simulated on the microscale to examine and quan fy how physical hydrologic parameters, such as rock-surface proper es and pore sizes, aff ect longitudinal and transverse NMR signals of real, complex media. Parameters studied were: magne c fi eld strength, rock suscep bility, pore coupling, and surface reac vity. Using the fi nite element method (FEM), simula ons of the spa al-and me-dependent magne za on evolu on in arbitrary pore geometries, diff usion regimes, and heterogeneous distribu ons of rock surface proper es, i.e., surface relaxivity, were compiled using an adapted generic diffusion model coupled with magne c gradient fi eld calcula ons. The numerical simula ons were validated using analy cal solu ons that are available for simple pore geometries. We observed a pore-size-dependent ra o of transverse T 2 and longitudinal T 1 relaxa on mes, and thus a pore-size-related and rock-suscep bility-dependent eff ec ve transverse surface relaxivity was deduced. This can be used to improve es mates of pore sizes and thus of permeability from transverse NMR relaxometry measurements. Simula ons of connected pore systems showed signifi cant infl uences of interpore coupling at hydrologically relevant pore sizes, e.g., fi ne sands. Depending on the dominant diff usion regime, the typically heterogeneous distribu on of surface relaxivi es in rocks and sediments, i.e., geological noise, can lead to a signifi cant underes ma on of derived pore sizes and thus of permeability.Abbrevia ons: DG, diff usion gradient; FEM, fi nite element method; FID, free induc on decay; NMR, nuclear magne c resonance.In petrophysical applica ons of NMR, the measured relaxation signals originate from the fl uid-fi lled pore space. Hence, in water-saturated rocks or sediments, the water content directly corresponds to the initial amplitude of the recorded NMR relaxation signal. Th e relaxation rate (longitudinal/transverse relaxation time ratio T 1 /T 2 ) depends on the pore size and physicochemical properties of the rock-fl uid interface (surface relaxivity) and on the concentration of paramagnetic ions or molecules (e.g., dissolved O 2 ) in the fl uid phases (bulk relaxivity) (e.g., Bryar et al., 2000; Keating and Knight, 2007). Th erefore, NMR can provide estimates of hydraulic parameters, i.e., pore size distribution or permeability. In this respect, there is a large variety of empirical or semiempirical relations (Kenyon et al., 1988;Dunn, 2002;Sen et al., 1990). Other relations frequently used as permeability estimators from NMR are based on geometrical considerations such as the Kozeny-Carman model (Carman, 1956, Ch. 1). Th ese permeability estimators apply the analytical relations between pore size and NMR relaxation behavior derived by Brownstein and Tarr (1979) for single unconnected pores. Th ese relations, however, are limited to simple, idealized pore models such as cylinders, spheres, or planar shapes. Furthermore, the infl uence...