Advancing the capabilities of refinery process models requires fundamental knowledge of hydrocarbon composition and processing behavior at the molecular level. In common practice, however, the main obstacle to reaching such a level of understanding is the difficulty to characterize the molecular composition of petroleum and its derived products with current analytical methods. A different approach is through the use of hydrocarbon composition modeling techniques to derive the molecular make up of petroleum fractions, thus enabling the development of molecular reaction models. The purpose of this study is to illustrate the application of this concept to model and simulate the vacuum gas oil hydrocracking process at the molecular level. At first the analytical characterization of the feed sample is transformed into a computational mixture of hydrocarbon molecules that is consistent with the chemistry of the actual oil sample. This molecular representation is then used as input to model the chemical transformations occurring in the hydrocracking reactor. The reaction network is organized in terms of reaction families, and reactivity parameters are modeled with quantitative structure/reactivity correlations. The developed model is tuned using experimental data obtained from a bench-scale hydrocracking reactor. Simulations showed that the model reproduces the product distribution by boiling range and hydrocarbon type, relevant product properties (e.g., API gravity), and process parameters such as hydrogen consumption and hydrocarbon vaporization, over a wide range of operating conditions.