The application of near infrared (NIR) spectroscopy for the prediction of the thermal decomposition behavior of lignocellulosic biomass (three types of woody biomass and three types of herbaceous biomass) was successfully performed along with statistical analysis. The thermal degradation behaviors of the woody and herbaceous biomass were different because of their different chemical compositions. Herbaceous biomass was degraded at lower temperature than woody biomass. The weight-loss profiles as a function of temperature were obtained by thermogravimetric analysis (TGA) at a heating rate of 25 degrees C/min under nitrogen gas. The weight-loss percentage at 10 temperatures in the range 150-600 degrees C was predicted by a wavelet partial least squares (PLS) model, which showed significantly better predictive performance than the ordinary PLS model. The results show that the data predicted by the wavelet PLS model was well fitted to the original data by TGA, in which the root-mean-square error in prediction values less than 5.5 suggested that NIR spectroscopy was applicable for rapid analysis to characterize the thermal decomposition behavior of lignocellulosic biomass for energy production. (C) 2009 Wiley Periodicals, Inc. J Appl Polym Sci 114: 3229-3234, 2009