The applicability of the Multi-wavelength Near-infrared sensor to analyze coal properties such as proximate analysis (moisture, ash, volatile matter, fixed carbon), ultimate analysis (carbon, hydrogen, nitrogen, oxygen, sulfur) and heating value is discussed. The most useful wavelengths (1,680, 1,942, 2,100, 2,180, 2,300 nm) for determining coal properties concentration were chosen by analyzing the NIR spectrum according to coal properties. Absorbances at the characteristic wavelength obtained from 128 mixed coal samples, which are using at a conventional thermal power plant, were correlated to the coal properties by using multiple regression analysis. The accuracy of coal analysis was examined by calculating the RMSEC (%), RMSEP (%), comparing the error with ASTM/ISO tolerance and performing paired Student's T-test. The result of on-line coal analysis for all moisture, volatile matter, fixed carbon, carbon, hydrogen and heating value is not different from that of ASTM/ISO traditional methods at 90% confidence level. The technology appears suitable for the determination of several coal prorperties. If calibrated periodically, this on-line analysis of coal properties is helpful to efficiently operate a coal fired power plant.