Despite recent focus on shale gas, hydrocarbon recovery from the ultraconfining and disordered porosity of organic matter in shales (kerogen) remains poorly understood. Key aspects such as the breakdown of hydrodynamics at the nanoscale and strong adsorption effects lead to unexplained non-Darcy behaviors. Here, molecular dynamics and statistical mechanics are used to elucidate hydrocarbon mixture transport through a realistic molecular model of kerogen [Bousige, C.; et al. Nat. Mater. 2016, 15, 576]. Owing to strong adsorption effects, velocity cross-correlations between the mixture components and between molecules of the same species are shown to be negligible. This allows estimation of each component permeance from its self-diffusivity, which can be obtained from single-component data. These permeances are found to scale with the reciprocal of the alkane length and decrease with the number of adsorbed molecules following a simple free volume theory, therefore allowing mixture transport prediction as a function of the amount of trapped fluid.