Near-infrared spectroscopy (NIR) is an efficient, low-cost sensing technology that has potential as an accurate biomass characterization method. The objective of this study is to develop NIR models in conjunction with chemometrics to determine high heating value (HHV) and elemental compositions of sorghum biomass. Partial least squares (PLS) regression and principle component regression (PCR) were used to develop calibration models with full and reduced wavelength regions. In general, models from reduced wavelength regions yielded higher calibration and prediction accuracies. Models to predict HHV, carbon, hydrogen, nitrogen, sulfur, and oxygen contents of sorghum biomass were well developed. HHV value, carbon, hydrogen, nitrogen, sulfur, and oxygen contents were predicted with R2 of 0.96, 0.96, 0.87, 0.86, 0.84, and 0.83 for validation sample sets, respectively. HHV and carbon content models had excellent prediction accuracy, whereas hydrogen, nitrogen, sulfur and oxygen models could provide reliable predictions. Those models provide good insight into the relationship between chemical bonds and HHV and elemental composition of sorghum biomass, allowing a rapid and accurate determination of HHV and elemental composition at low cost (from 200 to 1 USD) and reduced the time (from 100 to 1 min).