In this study, a theoretical methodology was developed for reconstructing the molecular composition of biocrudes based on a series of analytical property measurements with the purpose of facilitating detailed modeling of their reaction chemistry when they are processed into biofuels. The key characteristic of this methodology was treating biocrude molecules as a collection of recurring structural features (e.g., alkyl chains, aromatic rings, oxygen functional groups, etc.) that were assumed to be distributed in a statistical-like fashion. Biocrude molecules were systematically assembled following a building sequence where each structural feature was specified by means of Monte Carlo sampling of a designated statistical distribution function. The mixture of biocrude molecules was optimized by means of simulated annealing and entropy maximization so that it could match the analytical properties of the target sample. The molecular reconstruction model was put to the test with the light (<343 °C) and middle (343−460 °C) distillation fractions of a biocrude produced by hydrothermal liquefaction of forest biomass. The simulated mixtures, consisting of 1000 molecules each, matched reasonably well the actual fractions of the biocrude in density, elemental composition, boiling point distribution, and total carbonyl content, as well as in structural information obtained by nuclear magnetic resonance spectroscopy. The model's accuracy was better with the light fraction of the biocrude than it was with the more complex middle fraction. It was possible to extract insightful information about the chemical characteristics of the biocrude fractions from the simulations, such as the overall molecular weight distribution and the distributions of structural classes and oxygen functional groups.