Regarded as one of the most important welding methods, resistance spot welding is widely used within the automotive industry with mass production welding robots. Other industries are comprised of manual sheet fabrication, apparatus engineering, electrical appliances, and consumer household equipment like washing machines and tumble dryers. As a result of electrical fluctuations and the wear of electrodes, the quality of the joints in resistance spot welding cannot be guaranteed with ultrasonic waves or x-rays. That is up until now. And the current quality assurance systems involve random destructive testing, which is both costly and time consuming. Therefore the way forward was to create a new control system that would employ neural networks, whereupon the quality evaluation time had to be less than one-fifth of a second. Furthermore an improvement on the 1-2 percent non-detected defective welding spots had to be reduced to less than one percent, and so two prototype in-process quality assurance systems were investigated. It was soon realised that two separate applications were not necessary; one for electrical engineering and another for automobile manufacturing, and so the efforts were concentrated on just one model. With testing executed in both industrial sectors, the prototype compromised of a hardware module for data measurement, signal conditioning and data recording. A software module complements the hardware with its ability to measure fault detection, data pre-processing and online quality prediction. The prototype carried out tests under production conditions and can create multiple projects to observe various welding conditions, whilst welding time and sample rate can be built into several dialogues. It also has the ability to observe 256 welding programs and welding spots per welding object; which in turn should reduce existing quality control costs in the magnitude of 20-30 percent.