The high viscosity of heavy crude oil hinders its efficient production and transportation. Oil-soluble viscosity reducers (VRs) are crucial for ensuring the world's energy needs are met; however, the molecular mechanism underlying the reduction of viscosity remains unclear, which obstructs the selection of VRs for specific heavy oil reservoirs and further development of VRs. In this work, using molecular dynamics (MD) and quantum mechanics (QM) simulations, modeled heavy oils and VRs were employed to explore the molecular pathways of VRs to reduce heavy oil viscosity. Taking structural analyses and interaction energy analyses together, for a small asphaltene surrogate whose molecules interact with each other via hydrogen bonds, VR molecules were interspersed within the spatial hydrogen bond network, leading to a relaxation of network configuration. However, for another asphaltene surrogate that tends to aggregate through face-to-face stacking, the viscosity reducer could adsorb onto the asphaltenes, resulting in a shielding effect on asphaltene nanoaggregates from other heavy oil components. Thus, the viscosity reduction pathways should be chosen according to the molecular characteristics of the asphaltenes. These molecular insights and implications are expected to facilitate the future design and application of oil-soluble VRs.