Quantitative analysis of steel alloys was carried out using laser induced breakdown spectroscopy ( LIBS ) taking into account the complex matrix effects in steel alloy samples. The laser induced plasma was generated by a Q-switched Nd∶YAG laser operating at 1 064 nm with pulse width of 10 ns and repeated frequency of 10 Hz. The LIBS signal was coupled to the echelle spectrometer and recorded by a high sensitive ICCD detector. To get the best experimental conditions, some parameters, such as the detection delay, the CCD's integral gate width and the detecting position from the sample surface, were optimized. The experimental results showed that the optimum detection delay time was 1.5μs, the optimal CCD's integral gate width was 2 μs and the best detecting position was 1.5 mm below the alloy sample’s surface. The samples used in the experiments are ten standard steel alloy samples and two unknown steel alloy samples. The quantitative analysis was investigated with the optimum experimental parameters. Elements Cr and Ni in steel alloy samples were taken as the detection targets. The analysis was carried out with the methods based on conditional univariate quantitative analysis, multiple linear regression and partial least squares (PLS) respectively. It turned out that the correlation coefficients of calibration curves are not very high in the conditional univariate calibration method. The analysis results were obtained with the unsatisfied relative errors for the two predicted samples. So the conditional univariate quantitative analysis method can't effectively serve the quantitative analysis purpose for multi-components and complex matrix steel alloy samples. And with multiple linear regression method, the analysis accuracy was improved effectively. The method based on partial least squares (PLS) turned out to be the best method among all the three quantitative analysis methods applied. Based on PLS, the correlation coefficient of calibration curve for Cr is 0.981 and that for Ni is 0.995. The concentrations of Cr and Ni in two target samples were determined using PLS calibration method, and the relative errors for the two unknown steel alloy samples are lower than 6.62% and 1.49% respectively. The obtained results showed that in the quantitative analysis of steel alloys, the matrix effect would be reduced effectively and the quantitative analysis accuracy would be improved by the PLS calibration method.