We have outlined our approach for developing an integrated system for controlling centerline bead temperature in arc welding. An optical thermography system, based on an inexpensive and readily available video camera, was presented. Temperature measurements obtained with this system are within 10% of those measured by a calibrated two-color optical pyrometer. A low-order model describing the centerline temperature dynamics was developed. Two strategies for tuning a simple PI temperature controller were given. We first used off-line sensitivity-based tuning to optimize the controller gains. This method is adequate for systems whose dynamics are time invariant. For GMAW, however, the plant parameters can vary widely due to changes in plate thickness as well as other operating parameters. For this reason we applied a sensitivity-based pseudogradient adaptive algorithm for self-tuning the PI controller. The self-tuning controller is robust to changes in the operating conditions and is therefore more useful than a PI controller with fixed gains. It is also insensitive to the noise present in the GMAW system. The control design presented in this paper makes no attempt to counter the delay between the arc current command and centerline temperature output. Instead, the sample time was deliberately increased so that this delay could be ignored. To increase the bandwidth of the closed-loop system it is necessary to incorporate a Smith predictor or similar mechanism to increase the response time of the closed-loop system. We are currently investigating the design of a predictive controller.< <ETX xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">></ETX>