Near infrared (NIR) spectroscopy can be implemented in the evaluation of cellulose. The potential of NIR spectroscopy combined with multivariate analysis to evaluate moisture variation in pulp was studied. Samples of four pulp types were conditioned to different moisture levels. The samples were air dried in a controlled environment, at each 10 % moisture reduction the material was weighed and analyzed with the NIR spectrometer. Principal Component Analysis (PCA) and Partial Least Squares Regression (PLS-R) were applied to the spectral signatures and moisture values obtained during drying. Combining NIR spectroscopy with PLS-R, the moisture of the pulps under different conditions was estimated with R²p ranging from 0.89 to 0.98 for independent validation and root mean square error (RMSEP) ranging from 5.1 % to 18.3 %. The PLS-R models were applied to NIR spectra taken from other pulps and the estimates were consistent. The models showed robustness for monitoring pulps subjected to moisture variations.