We have developed a workflow for constructing realistic mesh-based magnetotelluric (MT) models from 3D geologic models. The routine is developed for unstructured meshes that adapt to the complex shapes of geologic bodies including 3D surfaces and volumes in realistic modeling scenarios. The methodology is applied to the complexly altered Lalor volcanogenic massive sulfide deposit in Manitoba, Canada. The host rock envelope of the Lalor deposit is compartmentalized into lithostratigraphic units leading to a watertight model. This model then is meshed into unstructured tetrahedral meshes suitable for synthetic geophysical modeling of the MT method. Subsequently, two 3D resistivity models are generated from wireline logs: (1) a host rock background model in which each tetrahedral cell is attributed with the average resistivity of each lithostratigraphic unit and (2) a heterogeneous background-ore model in which the resistivity values of the cells are resampled from a 3D curvilinear grid model, generated by computing sequential Gaussian simulations from the resistivity data for each unit of a 3D lithofacies model produced by categorical kriging. To calculate the synthetic response of this model for MT, a numerical-modeling code is developed based on solving the vector-scalar potential formulation of the electromagnetic diffusion equation using the finite-element method on unstructured meshes. After validating the numerical method for the Commemi test model, the MT response of the Lalor model is investigated. A reasonable agreement is observed between the survey field data and the data synthesized from our constructed heterogeneous model. Using an investigation of the inductive and galvanic parts, we conclude with the ideal frequency range for detecting the ore deposit. We also conclude with and visualize the importance of regional-scale alteration zones around the ore deposits and model inhomogeneities in boosting the detectability of the ore formations through feeding electrical currents as a result of galvanic field dominance at depth.