In evaluating the geometrical characteristics of mechanical part, cleverness may be added with the definition of an empirical model representing the “signature” left by the manufacturing process used to make the part. This manufacturing signature is the systematic pattern that characterizes all the features machined with that process. If such a model is available, it may be exploited to enhance geometrical inspection accuracy. In this paper, an approach for geometrical inspection of machined profiles is proposed. This approach consists in computing form deviations by reconstructing the actual profile using a frequency model of process signature. The method has been thoroughly investigated in different simulated scenarios and benefits in terms of improved accuracy are demonstrated. Within the paper, a case study, related to roundness of mechanical parts obtained by turning, is used. The relationships between the number of sampled points and fitting algorithms are also pointed out.