New correlations have been developed on dry ash-free basis using only proximate analysis or combination of both ultimate and proximate analysis. In this regard, 146 different biomass residues were selected for which consistent data on proximate analysis, ultimate analysis, and HHV are available in the open literature. The available values are converted into dry ash-free basis to develop linear and non-linear predictive models. Multilinear regression method is used to create the required correlations as HHV_daf=ƒ₁ (VM); HHV_daf =ƒ₂ (C); HHV_daf =ƒ₃ (H); HHV_daf =ƒ₄ (C, H); HHV_daf =ƒ₅ (VM, C); HHV_daf =ƒ₆ (VM, H); and HHV_daf =ƒ₇ (VM, C, H). The correlation accuracy was tested by considering various statistical parameters such as average absolute error (AAE), standard deviation of error (SD_err), coefficient of determination, and maximum absolute error. AAE is estimated in the range of 4.05 - 5.33 %. The standard deviation of error is lower than the previous analysis, which is about 1.2 MJ/kg. The coefficient of determination is estimated to be in the range of 0.03 - 0.42. It is found that the stated statistical parameters for the correlations developed on dry ash-free basis are much more favourable than those for the correlations available in the literature.