Near infrared spectroscopy technique combined with chemometrics methods was applied to predict crystallinity of Neosinocalamus affinins. Three improved partial least squares (PLS) methods, including interval partial least squares (iPLS), synergy interval partial least squares (siPLS) and backward interval partial least squares (biPLS), were used to find the most informative ranges and build models with better predictive quality based on multiplicative scatter correction spectra. And then the models were compared with PLS model which was developed on the whole wavelength range 350~2 500 nm. The results showed that the models built by the three improved PLS methods had higher predictive ability than that of PLS model, and the optimal model was obtained by siPLS method that separated the whole spectra into 30 intervals and combined three intervals. The siPLS model had correlation coefficient (R) of 0.88 and root mean standard error of prediction (RMSEP) of 0.011 7. Therefore, through selecting the effective wavelength range, siPLS method could accurately and rapidly predict crystallinity in Neosinocalamus affinins.