The reliability of a resistance spot welding (RSW) process is studied monitoring the quality of the corresponding welding points. Each welding point is uniquely represented by a specific resistance characteristic curve over time. Five learning resistance characteristic curves, the good quality of the related welding points was experimentally verified by means of a non-destructive technique, are selected as a reference to check the quality of welding points related to different process resistance characteristic curves. A first estimate of the quality of the welding point is made comparing the corresponding process resistance characteristic curve with the learning maximum, minimum and average resistance characteristic curves. Both good quality and defective (glued or squeezed) welding points are observed. In order to more correctly identify the quality level of each welding point, two different parameters comparing the related process resistance characteristic curve with the learning average resistance characteristic curve are applied. First, the residual resistance, as the difference at each instant of time between the two resistance characteristic curves, is considered. Then, the Euclidean distance, as the geometric distance at each instant of time between the two resistance characteristic curves, is adopted. Finally, the trend of the quality of the welding points as their number increases for welding electrodes with a fixed number of dressings is investigated.