A method is introduced about quantitative analysis of carbon in coal by LIBS (laser-induced breakdown spectroscopy) in the present paper, and it introduces data optimization technology and method based on spectral data integration, normalization and data screening processing to overcome the poor quality of precision in the analysis because of laser source energy fluctuation, self-absorption, sample surface roughness etc. It is showed that the standard deviation (SD) is less than 1.6% using the method to analyze C element content in coal, and this method also can be used for other element analysis for coal.