Laser-induced breakdown spectroscopy (LIBS) was used to calibrate the concentration of Cr in soils combined with Support Vector Machine. The Nd:YAG pulse laser with the wavelength of 1 064 nm was used as the excitation source. The grating spectrometer and the charge couple device were used as spectral separation device and the spectral detection device. The multiple linear regression and support vector machine were adopted to make quantitative analysis on Cr in soils respectively. The result indicate that the multiple linear regression can get more accurate informination of the spectral lines: the correlation coefficient is increased from 0.689 to 0.980 compared with conventional quantitative method. Thereofre, the the accuracy of quantitative analysis is increased. The slope about calibration curve with support vector machine of test set is nearly about 1 and the correlation coefficient is 0.998, the relative errors for the test set all are lower than 2.57%, the quantitative analysis results about support vector machine are better than the results combined with the conventional quantitative method and the multiple linear regression. The support vector machine can correct the matrix effect and improve the accuracy of prediction on the concentration of Cr in soil.