In order to improve heavy oil characterizations for modeling crude oils in refinery processes, an in-house Deep Vacuum Fractionation Apparatus (DVFA) was recently modified to provide reproducible extended distillation data for heavy oils. In addition, an interconversion method was developed to obtain the Normal Boiling Points (NBP) of the distillation fractions. In the current contribution, the physical properties of the distillation cuts collected using the DVFA (the development dataset) are measured and used to evaluate existing correlations and develop new correlations to predict normal boiling point (NBP), specific gravity (SG), and molecular weight (MW) of heavy distillation cuts. The new correlations were also tested on an independent test dataset obtained from the literature. A modified version of Soreide's correlation was proposed for either NBP or MW. The correlation decreased the average absolute relative deviations (AARD) of the NBP from 3.0% to 2.0%, and of the MW from 7.8% to 5.3% for the development dataset and from 2.5% to 2.3% for the test dataset. Similar improvements were obtained for the predicted MW. A new correlation was proposed to predict SG from the H/C ratio and MW, which decreased the AARD from 1.7% to 0.8% for the development dataset and from 3.1% to 1.4% for the test dataset. Of greater significance, the new correlations improved the trend in the predicted properties toward higher boiling cuts and therefore, are expected to provide more accurate estimates of extrapolated properties for heavy oils. (C) 2016 Elsevier Ltd. All rights reserved.