The objective of this study was to investigate whether it is possible to use a lateral (profile) photograph to determine the underlying skeletal Class and which reference points of the angle of convexity are most suitable for this purpose. Profile photographs and lateral cephalographs included in the baseline data for 180 orthodontic patients were retrospectively evaluated. The subjects were assigned to skeletal Classes based on Wits values obtained by radiolographic analysis. The Class I subjects were 58 patients (22 males, 36 females) with an average age of 13.63 +/- 2.1 years, the Class II subjects 60 patients (37 males, 23 females) with an average age of 13.60 +/- 2.6 years, and the Class III subjects 62 patients (28 males, 34 females) with an average age of 11.65 +/- 3.3 years. The angles measured were A'OrB' (=POrA'-POrA'), A'N'B', and the angle of convexity with its variants (N'SnPog', N'A'Pog', TrSnPog', TrA'Pog', Gl'SnPog', and Gl'A'Pog'). These angles were statistically evaluated using a two-sided t-test and linear discriminant analysis. Class II and Class III subjects exhibited highly significant differences (P < 0.001) for all angles. Class I and Class III exhibited highly significant differences (P < 0.001) for almost all angles, and significant differences for A'N'B' (P < 0.05). Class I and Class II differed significantly (P < 0.05) only for some angles (N'SnPog', TrA'Pog', Gl'SnPog', and Gl'A'Pog'). The error within the linear discriminant analysis was smallest for N'SnPog', GlA'Pog', and TrA'Pog' angles. However, the method error according to Dahlberg yielded rather high values for all angles (1.07-1.17 degrees). Discrimination between skeletal Class I and Class III was easier than that between Class I and Class II. One of the reasons may be that the subclasses division I and division II were not distinguished within the Class II subjects.