Abstract: The power signal in the welding process was employed for the thickness of 0.4 mm TC2 titanium alloy in order to monitor and predict the nugget diameter. The voltage between the upper and lower electrode was obtained using the twisted pair cable and the welding current was acquired based on the Rogowski coil. On this basis, the welding power curve during the resistance spot welding process was obtained. The changes of the welding power curve during the welding process were analyzed, and the characteristic values that characterize the welding quality were extracted. The results show that the welding power curve was not only consistent with the change of the dynamic resistance curve, but also more related to the input of welding heat input. The correlation coefficient between welding heat input, accurately extracted from welding power curve, and nugget diameter is as high as 0.9. A mathematical model for predicting the nugget diameter with the eigenvalues of the dynamic power curve can be obtained through the powerful nonlinear mapping ability of the kriging algorithm. The maximum relative error between the predicted value and the actual value of the model was about 8.7%, which can effectively predict and monitor the welding quality of TC2 resistance spot welding with a thickness of 0.4 mm in real time.