Purpose: Functional relationships between occlusal errors and articulator settings have been the subject of numerous investigations. It is, however, not known how frequently errors of a certain size occur under practical conditions. This study attempted to estimate the frequency of occlusal errors that can be expected with the different levels of registration and articulator adjustment that are currently used. Materials and Methods: In 57 asymptomatic subjects, individual articulation parameters, including the condylar and Bennett angles, the spatial relation of dental arches, and the intercondylar distance, were determined experimentally. Tooth-guided lateral movements of the subjects' casts were simulated in a virtual articulator that was programmed to the individual parameters of each person. With respect to this reference setting, occlusal errors that would emerge with different modes of semiindividual adjustment were calculated. The intercondylar distance thereby remained fixed at 110 mm. Results: With complete mean value setting, occlusal errors would exceed 200 µm at the second molar in 16% of the subjects and 300 µm in 6% of the subjects. Individual facebow registration of condylar angles and spatial relations would reduce these rates to 13% (200 µm) and 3% (300 µm). With additional setting of Bennett angles, occlusal errors would exceed the mentioned limits in no more than 1.6% and 0.1% of cases, respectively. This extensive mode of adjustment became, however, increasingly ineffective with higher demands for occlusal accuracy. Conclusion: Complete mean value setting is associated with a relatively low risk of occlusal errors exceeding tolerance limits that are widely accepted in practice. Compared to mean value setting, facebow registration of condylar angle and relations yields no profitable improvement of occlusal accuracy.