The aim of the present long-term follow-up study of orthodontically treated patients was to analyze postretention changes in arch width dimension and to isolate factors that may serve as predictors of long-term prognosis. Pretreatment, end-of-treatment, and postretention (at least 10 years) models of 226 cases with different malocclusions were used to measure intercanine and intermolar width, arch length, sum of the mesiodistal dimension of the incisors, irregularity index, crowding, molar and canine relationship, overjet, and overbite. To assess the influence of initial and end-of-treatment alignment, kind of treatment (extraction versus nonextraction) and the amount of expansion in postretention stability, the sample was divided into different subgroups. The findings indicate that postretention arch width relapse occurred more frequently in the upper intermolar (25.8%) and lower intercanine region (23.9%) than in the lower intermolar (19.0%) and upper intercanine (13.8%) region. Pretreatment and posttreatment alignment as well as the kind of treatment and the amount of expansion were found to be influencing factors. The study concludes by proposing a reassessment of the definition of stability. The influence of the pretreatment anomaly, kind of treatment, amount of expansion, and posttreatment alignment on long-term stability should be recognized. Patients should be apprised of treatment limitations before treatments.