Class III malocclusion is one of the most difficult anomalies to understand. Because not all Class III patients are candidates for surgical correction, patient assessment and selection remain main issues in diagnosis and treatment planning. The purpose of this study was to separate Class III patients who can be properly treated orthodontically from those who require orthognathic surgery. A large sample size was a necessary to obtain a sufficiently robust model. Thus, a multicentric study design was chosen (Orthodontic Departments of the Universities of Frankfurt, Heidelberg, and Wurzburg, Germany). The cephalograms of 175 adult patients with Class III malocclusions were analyzed. The orthodontic group comprised 87 patients, and the surgery group, 88 patients. Twenty linear, proportional, and angular measurements were made. Stepwise discriminant analysis was applied to identify the dentoskeletal variables that best separate the groups. The discriminant function model was highly significant (P <.0001); 92% of the patients were correctly classified. The following variables were extracted: Wits appraisal, length of the anterior cranial base, maxillary/mandibular (M/M) ratio, and lower gonial angle. The resulting equation was: Individual score = -1.805 + 0.209 . Wits + 0.044 . SN + 5.689 . M/M ratio - 0.056 . Go(lower). By means of discriminant analysis, correct classification of adult Class III malocclusion patients succeeded to a very high degree. Of all the variables, the Wits appraisal was the most decisive parameter.