BACKGROUNDIn the bioethanol production process it is necessary to know the physical and chemical characteristics of pretreated material which may affect the enzymatic hydrolysis as well as the carbohydrates potential for fermentation. In the present study, prediction models were developed with Fourier transformed near infrared (FT-NIR) spectroscopy and multivariate analysis for the cellulose crystallinity index (CrI), saccharification and ethanol yield, and glucan, hemicellulose and lignin content. Prediction models were generated using Eucalyptus globulus and Pinus radiata samples from different pretreatments such as bio pretreatment, kraft process, steam explosion, organosolv and biorganosolv process. RESULTSThe calibration models obtained by partial least-squares (PLS), presented high coefficients of calibration and validation determination. The predictive ability of the calibration models developed was evaluated in a set of external samples. The properties were predicted with a prediction error (RMSEP) of 0.02, 1.4gL(-1), 2.0%, 0.9%, 1.0% and 1.9% for CrI, ethanol yield, saccharification, lignin, hemicellulose and glucan content, respectively. CONCLUSIONThese results suggest that it is possible to construct models for these properties with a database involving pulps from different pretreatments. NIR spectroscopy predictions provide useful and economic tools for rapid analysis of the large number of samples required in the assessment of bioethanol production processes. (c) 2014 Society of Chemical Industry