The width of spectral line acquired by laser induced breakdown spectroscopy (LIBS) is narrow. Spectrometer exists wavelength shift as the temperature of external environment is changing. The discrete digital spectral data is influnced by noise, spectral overlap, and continued background emission. The signal has distortion and deformation. The method of spectral line recognition in current spectrometer software is maily the principle of proximity which results in high error rate and depends on people′s observation. Considering this problem, based on the study of characteristic of LIBS, A method of correlative analysis with an alterable and scannable window (CAASW) is presented in spectral automatically recognition of laser-induced breakdown spectroscopy. In experiment, using soil sample, the process of the method is analyzed and the results are evaluated. Compared with the spectral recognition software of spectrometer, the CAASW is more exact and less time spending, and it can recognize spectral lines automatically.