The feasibility of using computer-based parameters for quantifying mandibular asymmetry was investigated. Four methods of calculating asymmetry were used, based on the digitized facial photographs of three groups of patients: those with no observable asymmetry, a group with mild asymmetry, and a group presenting for orthognathic surgery. Three of the methods involved right/left difference ratios, namely, area, perimeter length, and compactness. The fourth, moment ratio (centre of area), was expressed as a percentage. Repeatability of both photography and digitization proved satisfactory, the standard deviation of the differences between repeated photographs being 0.016 and 0.014 for area and compactness ratios, respectively. Area, perimeter, and compactness successfully discriminated between the three groups. For area, median ratios (deviations from 1.00) for the 'normal', 'mild asymmetry', and 'surgical' groups were 0.015, 0.030, and 0.078, respectively. Those patients in the surgical group for whom asymmetry correction had been the main reason for surgery recorded higher asymmetry scores than the other subjects in that group. Moment ratio did not adequately distinguish between the three groups. Better repeatability for digitization was found when a baseline involving the ear insertions was used, than either the outer or inner canthi of the eyes. The potential uses of this approach are presented in relation to clinically relevant mild asymmetry, as well as auditing the outcome of surgical correction.