Natural rubber (NR) blends are widely used in many industries because of their excellent integrated properties. However, a simple, easily operational, nondestructive, and accurate method for their quantitative analysis remains as a challenge. This has been always an important issue in the related industries, particularly for their daily quality control tests. One main reason is that NR ingredients vary according to their geographical origin and the harvest time, which renders it hard to set up a versatile analytical protocol for all NRs. Another reason is owing to the defects of the established methods themselves as having been revealed in those relying on TGA, Py-GC/MS, FTIR, and ATR-FTIR. In this study, a simple and feasible method based on near infrared spectroscopy combined with chemometric is proposed to solve this problem for the first time. NR/SBR (styrene-butadiene rubber) rubber blend, the most widely used NR blend, is selected as a typical research subject. Spectral calibration region, factor, and several different pretreatment methods are applied on the spectra data to optimize calibration models. The result shows the optimized calibration model provides a good accuracy (0.135 wt %), intraday precision (0.121 wt %) and interday precision (0.132 wt %) for 3 months. V C 2014 Wiley Periodicals, Inc. J. Appl. Polym. Sci. 2015, 132, 41423.