Knowledge of the molecular weight distribution is important for rationalizing coal behavior. While many analytical approaches generate average data, inclusion of coal's inherent structural diversity would improve molecular representations of coal and their usefulness. The molecular weight distribution of Pocahontas No. 3 coal was estimated based on a new approach coupling HRTEM lattice fringe image data and laser desorption ionization mass spectra (LDIMS), constrained by elemental and NMR data. Assuming a shape for the large aromatic coal molecules allows the determination of the aromatic raft size distribution, and prediction of the molecular weight distribution from the HRTEM lattice fringe image analyses. Similar-shaped molecular weight profiles were obtained from these different techniques. Both distributions showed a sharp rise, fall and long tail, with the HRTEM profile being shifted to a lower mass in comparison to the LDIMS data. The mean molecular weight of an aromatic raft, 289 Da, was similar to 299 Da a value reported from NMR data. Cross-linking the fringes generated a diverse network structure of aromatic clusters with a reasonable aromatic H/C ratio and a molecular weight distribution within the appropriate ranges from laser desorption data. A rationale for molecular diversity determination, necessary for large-scale molecular modeling, for a high-rank coal is proposed. (C) 2009 Elsevier Ltd. All rights reserved.