A lot of information reflecting laser welding process are involved in laser-induced plasma. The electrical signal and spectral signal from laser-induced plasma are detected by a synchronous electrical and spectral information acquisition system and the signals are compared. And then the relationship between electrical signal and plasma temperature is illustrated. The electrical signals of the induced-plasma for stainless steel A304、stainless steel 430 and carbon steel Q235 in laser surface auto genous welding are detected in real-time and analyzed by the probability density distribution. The analysis of the standard deviation of probability density distribution of the electrical signal from the strainless steel A304 laser welding with the most obvious deep penetration characteristics is presented. The results indicate that the temperature variation of laser induced-plasma can be real-timely detected by the electrical signal, and laser welding mode can be identifed by the probability density analysis of the electrical signal of the laser induced-plasma and the laser welding mode of stainless steel A304 can be determined by the standard deviation analysis.