The prediction of weld nugget diameter has been a primary objective in the development of resistance spot welding monitoring techniques. The prediction algorithms must be able to work with the wide range of process variations commonly encountered in a production environment. Welding experiments simulating process variations were conducted on AA5754 sheet with structural adhesive using AC welding equipment. The parameters and simulated production conditions were extremely wide ranging, and included welding sections of actual car components. Results from the work have demonstrated that by using suitable electrical signals sampled from the process, robust models can be designed that are capable of making good diameter predictions under a wide range of welding conditions.