Muon tomography is applied to realistic density models of a steam-assisted gravity drainage (SAGD) reservoir at 1.25 and 5 years after initial reservoir production. Forward models of muon count and opacity based on the density models are computed, as well as inverse models of the synthetic muon observations for various simulated detector arrays. The results demonstrate that both phases of reservoir development, namely the rising phase and the spreading phase, can be resolved by muon detectors placed 30 m below the bitumen reservoir at 230 m total vertical depth. The total mass change in the reservoir was recovered from the inversion model and differs from the true mass change by 20%–29%. The spatial distribution of density change shows very good agreement in the horizontal direction, while the vertical is less well constrained in this modeled sensor array configuration. The inverse models provide improved insights into reservoir depletion patterns and indicate muon tomography to be an applicable tool for continuous reservoir monitoring. The numerical modeling approach developed herein is able to model a wide range of SAGD reservoir geometries and detector arrays toward planning of optimized monitoring solutions.