The optimal conditions for the automation of GMAW are established by the minimization of welding cost under the constraints assuring the weldability. A weighting factor, W is introduced into the cost function intentionally to restrict the welding speed with which errors in welding process is related. The effects of welding parameters such as welding current, voltage, and speed on the penetration, the excessive reinforcement and the deposition efficiency is derived by experiment to give some constraints. Gradient projection method is used for the numerical calculation of optimal conditions. As a result of the numerical calculation, it is known that (1) welding costs are influenced mainly by welding speed, and such an influence become weak at the high degree of welding automation (i.e. as the operating factor approaches to 1.0), and (2) to the extend of W≤0.02, it is reasonable to introduce the weighting factor to the cost function and the cost saving by the welding automation is much more effective in the low speed welding.