Abstract: A new method of qualitative inspection of textiles by Near Infrared spectroscopy technology was put forward in order to avoid the time consuming and tedious work and pollution in the traditional inspection. Qualitative clustering analysis of 75 textile samples, including cotton, terylene, cotton-terylene and cotton-polyurethane, was done via Near Infrared spectroscopy technology. Eigenvalue spectrum was extracted with principal component regression. P.C.Mahalanobis was used in classification analysis. The outcome proved that it was feasible to inspect the components of textiles via Near Infrared spectroscopy technology. Spectrum translation is the primary noise in textile Near Infrared spectrum, which was shown by comparing different spectrum pre-processing methods. Some guidance is presented for further raising the accuracy of the model.