The adsorption of nanoparticles onto mineral surfaces is a major limitation for applications that require long transport distances, such as enhanced oil recovery. This study investigates silica nanoparticle transport and adsorption in long granular columns, with emphasis on the adsorption onto carbonate substrates, given the fact that carbonate reservoirs host more than 60% of the world's recoverable oil. The grain-scale particle-mineral interactions are characterized by zeta potential measurements. Ionic strength (especially potential-determining ions: Ca2+, Mg2+, CO32-, etc.) inherently influences the zeta potential of carbonates. Derjaguin-Landau-Verwey-Overbeek analyses show that low surface potential and high ionic concentration inhibit the electrostatic double-layer repulsion and lower the energy barrier of adsorption. Adsorption column experiments simulate a variety of fluid chemistry conditions: pH, ionic concentration, and ion type. Alkaline and low-salinity conditions favor silica nanoparticles transport in carbonate reservoirs. Both scanning electron microscopy images and adsorption mass analyses suggest that the adsorption of nanoparticles onto carbonate substrates is multilayered. A two-term adsorption model adequately captures the instantaneous adsorption and the subsequent kinetic adsorption. The instantaneous adsorption constant delays particle transport, and the kinetic adsorption rate determines the concentration profile of nanoparticles along the reservoir at the steady state. High advection velocity and low adsorption rate k(1) are required to deliver high nanoparticle concentration to the far field in the reservoir.