It is important to realize rapid detection of straw content in co-firing biomass-coal blends for drawing up reasonable precept about biomass co-firing power generation subsidy. The original samples were consisted of 81 straws and 9 coals collected from different regions and different varieties in China. 90 blends samples of coal and straw (1%～30%) were prepared after comminution and separated into a calibration set (60 samples) and an independent validation set (30 samples). Spectra were scanned by FT-NIR spectrometer. The interval partial least squares (iPLS) and the genetic algorithm (GA) were used for wavelength selection. Quantitative analysis models for straw content were established by partial least squares (PLS). The results showed that data points for modelling decreased from 3001 to 33, determination coefficient, standard deviation of prediction (SEP) and ratio of performance to standard deviation (RPD) in validation were 0.89, 2.87% and 3.06 by GA-PLS method, respectively. It is concluded that NIRS with GA-PLS was feasible for fast quantitative analysis of the straw content in co-firing biomass-coal blends. 