Manganese (Mn) is one of the indispensable micronutrients for plants. The concentration of Mn in soil was determined with laser-induced breakdown spectroscopy (LIBS) in this paper. Forty-six soil samples were used to obtain the spectrum data of LIBS. The characteristic line of 403.1 nm was selected as the analytical line of Mn. The calibration curve of the net intensity of Mn line at 403.1 nm versus the corresponding concentration was constructed withthe correlation coefficient 0.78. The results showed that, due to the complex chemical compositions of soil samples, the calibration method influenced by the matrix effect seriously. It should utilize more information of LIBS spectra to construct the multivariate nonlinear calibration method, which can reduce the matrix effect and improve the measurement accuracy of LIBS. With multivariate nonlinear calibration method, the interference effect of C and Fe elements on the concentration of Mn was considered. Compared with calibration method, the correlation coefficient of the prediction concentration of LIBS with reference concentration was 0.97, and the relative measured error was in the range 3.2%~10.3%,Therefore, the measurement accuracy of LIBS was increased. The experimental results demonstrate that LIBS combined with multivariate nonlinear calibration method can be used to quantitative analysis of Mn in soil.