Abstract: To apply the micro-magnetic testing method to the induction hardened case depth evaluation, an integrated sensor was designed to synchronously detect the tangential magnetic field strength signal, the hysteresis curves and the Barkhausen noise. A total of eight features or parameters were extracted from the above mentioned three kinds of signals for induction hardened case depth characterization in steel rods. Stepwise regression method was employed to help select the parameters with a significant level lower than 0.07. Finally, the coercive force H_c, the amplitude of the three harmonics (A₃), the distortion factor (K) of tangential magnetic field strength signal, and the parameter H_cm of the butterfly-like Barkhausen noise envelop were selected as the input of a multivariable linear regression model. The established linear regression model had an averaged prediction error of 3.87%. The developed experimental set-up together with the signal analysis method is expected to be applied to quantitative measurement of the induction hardened case depth in rod-like ferromagnetic components.