In the present study， the rapid prediction of wood modulus of elasticity (MOE) of Eucalyptus pellita by near infrared (NIR) spectroscopy is described. Fast Fourier transform (FFT) and conventional mechanical testing methods were used to measure modulus of elasticity of small clear wood samples of Eucalyptus pellita. After collecting the near-infrared reflectance spectra of each sample from radial and tangential faces， the NIR spectra were preprocessed with the second-derivative methods, and regression models were built between 410 to 2 480 nm. The calibration models were established using two thirds of whole samples with the partial least squares method, and validation models were developed on an independent set (one third of whole samples). The analysis results showed that high correlation coefficients were obtained between the laboratory-determined MOE values and NIR prediction values of Eucalyptus pellita. The correlation coefficients of prediction model for MOE were 0.93 and 0.81， and RPD were 2.70 and 1.71. NIR analysis technique can realize the rapid prediction of the MOE of small clear wood samples of Eucalyptus pellita.