Abstract：There are two problems in the graphic method and the Least Quares(LS) method in determining the shear strength of rock. (1) The experimental data should be highly correlated, and separate data and anomalous data are not suitable. (2) The effect of anomalous values in test data is markedly shown because of the quadratic sum of residual errors in the least squares method. This paper proposes a shear strength parameter(Robust) regression analysis method. This method can be applied to analyze experimental data with poor correlation, and with anomalous values, and the effects of discrete and abnormal positioning can be weakened to improve the robustness and reliability of estimated parameters. In order to reduce the effect of anomalous values in the test data, the sum of residual absolute value is used instead of the quadratic sum of residual errors in the robust regression method, then the quadratic terms of anomalous value could be avoided, and the effect of anomalous values could be reduced. It is shown that the results of the two methods are the same when the test data is good, but the results of the robust regression method are better than those obtained by the least squares method when there are anomalous values in the test data.