Background and aim: Contemporary orthodontic and surgical treatment goals are primarily focussed on achieving optimal aesthetic soft tissue outcomes in three dimensions. It is important, therefore, to establish valid three-dimensional normative models to assist in clinical decision-making. Ideally, such models should be customised to a patient’s individual facial proportions. The aim of this study was to establish the most pleasing computer generated 3D facial form using a community-based sample population. Methods: Three-dimensional facial surface data (3dMDface) were obtained from 375 young adult Caucasians (195 males and 180 females, all approximately 22 years old) without craniofacial anomalies, all of whom were participants in The Raine Study in Western Australia with participants from Generation 2. These data were used to generate seven faces that represented the variations in convexity distributed evenly around an average. The faces were subsequently rated by orthodontists, oral surgeons, plastic surgeons, dentists and laypeople for attractiveness. Results and conclusion: Age, sex and occupation did not influence the preference among the various faces. The average face was rated as the most attractive. For males, a slightly concave profile and for females a slightly convex profile was preferred. The present study suggested that orthodontic/surgical treatment of Caucasians should be directed towards achieving an average facial form.