Abstract: The yield strength and the tensile strength of DP590 steel strip were nondestructively and quantitatively predicted based on the micromagnetic testing method. Ten specimens were prepared and two-stage annealing method was employed to change the yield and tensile strength of the specimens. A multi-functional sensor was used to simultaneously detect the tangential magnetic field and magnetic Barkhausen noise in the surface of the tested specimens. Multiple feature parameters representing the magnetic properties of the tested material were extracted from the obtained signals. The coefficient of variation analysis method was applied to evaluate the repeatability performance of the experimental set-up with respect to the measured magnetic parameters. Comparative studies were conducted to investigate the accuracy of multivariate linear and quadratic models in predicting the yield and tensile strength of the specimens. The results show that the multivariate quadratic model has better performances than the multivariate linear model. The coefficient of determination of the multivariate quadratic model is higher than 0.97. The prediction errors of the multivariate quadratic model to both the yield and tensile strength are less than 3%.