Rapid population growth, continuous climate change scenarios, and diminishing fossil fuel reserves make it crucial to identify sustainable alternative energy sources. Lignocellulosic biomass represents the largest quantity of biomass on earth and is an attractive source for sustainable and environmentally friendly biofuels and other bio-based chemicals. However, fast and accurate biomass characterization is essential for the biofuel industry. The conventional chemical analysis of biomass is expensive, tedious, and requires skilled personnel. This study assessed the suitability of near-infrared spectroscopy (NIRS) for high-throughput characterization of sorghum biomass composition such as moisture, ash, and acid-insoluble lignin (AIL) for biofuel production. Spectral data were acquired in the 867-2535 nm range, and their corresponding biomass compositions were determined following the laboratory analytical procedures developed by National Renewable Energy Laboratory. Standard normal variate (SNV), multiplicative signal correction (MSC), and Savitzky-Golay (SG) spectral pre-processing techniques were employed before developing prediction models using the partial least square regression (PLSR). The PLSR model performance was evaluated by an independent validation set. The developed predictive models with full spectra show acceptable model statistics for the selected components. Some dominant informative spectral wavelengths were used to develop more effective prediction models. This study revealed that it is possible to develop a low-cost, portable, and real-time NIRS technique to predict the moisture, ash, and AIL content of lignocellulosic biomass for biofuel production.