To investigate the feasibility of fast detection of the elements of N, C, H, S and O of straw biomass by using the near-infrared spectroscopy (NIRS) technology, 199 straw samples have been collected and prepared, including the straw of rice, wheat, canola and corn. Near-infrared diffuse reflectance spectroscopy combined with PLS chemometric algorithms has been used to compare the calibration effect with different spectral pretreatment methods in 7400-5550 cm-1 wavelength range, and the optimal calibration analysis models for N, C, H, S and O element of straw biomass have been established, then the independent samples of validation set were used to validate the model. The validation results show that the established quantitative analysis model for N element can be used in practical detecting; the established quantitative analysis model for O element can be used in practical estimation; the quantitative analysis model of C element by using near-infrared technology is feasible, but the model needs to be further optimized; H and S element can not be quantitatively analyzed by using NIRS technology.