The accuracy and precision of certain external cephalometric measurements were investigated on a subject presenting optimal conditions, using ten operators. Due to difficulties in identifying landmarks and tissue compressibility, the standard deviation of a measurement for ten replications ranged from 0 . 3 to 1 . 7 mm, while the "between operators" standard deviation was no better than 2 . 0 mm. Measurement accuracy was also poor: that of primary interest, subnasale-gnathion, had a bias of the order of 7 mm, judged by a radiographic determination. A minimum denture success rate of 97% is equated with a standard deviation of 1 mm in the design freeway space. A recently published regression equation for predicting occlusal vertical dimension has a minimum error of prediction of 5 . 8 mm and thus is considered an inadequate predictive tool. The errors demonstrated in facial measurements suggest that using formulae to predict occlusal vertical dimension is unlikely to produce clinically acceptable results.