When biomass is thermally treated, the enrichment of carbon in the remaining "green coal" is correlated with the temperature and duration. Other properties related to the energetic properties of the torrefied biomass are closely related to chemical modifications and correlated to the material mass loss occurring during the thermal degradation. The possibility of using near infrared spectrometry has been investigated to predict the mass loss ofPinus sylvestriswood torrefied at temperatures ranging from 220 degrees C to 300 degrees C with durations varying from 1 minute to 10 hours. A first mass loss prediction model (NIR-260) associated with the mean torrefaction temperature of 260 degrees C was developed, and appeared suitable only for this temperature due to specific chemical reactions rate. A second model (NIRS-All), using all available data was constructed and showed an accurate mass loss prediction, for both low (220 degrees C) and high temperatures (300 degrees C). The main differences between NIRS-260 and NIRS-All models are mainly attributed to the thermal modification of hemicelluloses and cellulose fractions occurred during the wood torrefaction. The results showed near infrared spectrometry combined with multivariate calibration modeling have potential utility in an industrial context as a standardized continuous method to figure out the mass loss of biomass during torrefaction by a rapid characterization. Novelty Statement The novelty concerns the use of the Near Infrared Spectrometry (NIRS) combined with multivariate calibration modeling as a standardized method for determining the mass loss biomass during torrefaction by a rapid and nondestructive characterization. A model was constructed and showed an accurate mass loss prediction, for both low (220 degrees C) and high temperatures (300 degrees C). Near infrared spectrometry have potential utility in an industrial context as a standardized continuous method.