Fifty seemingly well-treated orthodontic cases were studied by means of pretreatment, posttreatment, and postretention dental casts, lateral cephalograms, and other orthodontic records. The sample was restricted to cases exhibiting anteroposterior and/or vertical dysplasia as revealed by pretreatment dental casts. The sample was divided into a stable group and a relapse group. Each group contained twenty-five cases. A double-blind design was used. The raw data were analyzed by the stepwise discriminant analysis and by the multivariate analysis of variance. On the basis of the results obtained from this study, the following conclusions can be drawn: 1. In seemingly well-treated orthodontic cases, relapse or stability can neither be predicted nor judged from one set of records alone. 2. Relapse or stability of an orthodontic case can be predicted by comparing the posttreatment variables with the pretreatment variables. 3. Relapse or stability of an orthodontic case can be judged by comparing the postretention variables to the posttreatment variables. 4. The PP-GoGn angle and the mandibular intercanine width are the two most important variables associated with orthodontic relapse. 5. Changing the PP-GoGn angle, either by treatment or by growth, was associated with relapse. In other words, changes in the PP-GoGn angle tended to be unstable. This suggests that decreasing the PP-GoGn angle should be avoided. 6. In both the stable and the relapse groups, the mandibular intercanine width decreased postretention. This decrease was associated more with the relapse group than with the stable group. 7. The mandibular intercanine width tended to relapse toward its original pretreatment value. This suggests that, at the end of active treatment, the mandibular intercanine width should be maintained as originally presented. 8. There was no significant interaction between orthodontic relapse (or stability) and Angle Class I and II cases. 9. There was no significant interaction between relapse (or stability), of an orthodontic case, and the sex of the patient. 10. There was no significant interaction between orthodontic relapse (or stability) and whether or not extraction was included as a part of the mechanotherapy. Thus, the pretreatment deep overbite is not necessarily a contraindication to extraction.