Navigation-assisted surgery is the gold standard for complex reconstructive procedures of the midface and facial skeleton, and artificial and anatomical landmarks are often used for reference. The correct identification of these landmarks before surgery is crucial for the accuracy of the navigation system. This study aimed to investigate the human errors in reference point placement. This retrospective study investigated 228 reference-point positions in 51 cases where navigation was utilized. The discrepancies between the actual reference point-position and manually planned preoperative reference points were quantified using Brainlab iPlanCMF 3.0.6. The referencing methods used in these cases included dental registration splints, osteosynthesis materials, anatomical landmarks, and combinations of these methods. The average discrepancy in the actual and manually planned reference points was 0.29 ± 0.41 mm. The use of anatomical landmarks demonstrated a significantly lower deviation (p < 0.05), although the differences between the errors in reference-point placement using dental registration splints, osteosynthesis materials, or combinations of these methods were not statistically significant. The frequency of misplacement of reference points was significantly higher than expected. These errors might have been caused by human bias during manual placement of the points or intraoperative difficulties caused by extensive metal artifacts. Thus, we postulate that the surgical personnel involved in planning navigation-assisted surgery should undergo intensive training. The development of new referencing methods that are less susceptible to these causes of error might help overcome human bias.