Legg-Calvé-Perthes disease (LCPD) often produces a residual deformity, typically consistent with coxa magna, coxa plana, and ellipsoidal shape. Depending on the degree of asphericity and flatness, this morphology was classified by Stulberg in stages III and IV. Thus far, few studies have investigated physeal injury as an etiological cause or evaluated its progressive profile throughout Waldenström's reossification stage and the remodelling stage. In this study, we analysed the ellipsoidal process of the femoral head. This was a retrospective control case study involving 83 unoperated hips with LCPD and Stulberg stages III and IV outcome. The data were compared with those obtained for 49 healthy contralateral hips (control). The Ellipsoidal Index, the presence of a double epiphyseal reossification nucleus, physeal narrowing, intraphyseal angle, epiphyseal height, diameter of the head, and Reimer's Index were determined. Measurements were performed at four-time points: the year the reossification stage was initiated, the final growth stage, and two equally spaced time points in between. The Ellipsoidal Index gradually increased throughout the course of the disease from 1.6 in the initial reossification stage to 2.0 at the end of growth. In the control cases, this value was consistently 1.4. More ellipsoidal deformity was observed in Stulberg stage IV versus Stulberg stage III patients (P < 0.05). Moreover, there was a direct link between a high Ellipsoidal Index and the appearance of a double reossification nucleus, a physeal narrowing in the area underlying the anterosuperior nucleus, intraphyseal angle. Reimer's Index showed a gradual extrusion from baseline to the end of growth (26.1 versus 31.8, respectively; P < 0.05). The ellipsoidal process of the femoral head occurs gradually throughout the reossification and remodelling stages. This was linked to the appearance of a double epiphyseal nucleus, gradual extrusion, an angulated physis appearance, an asymmetrical narrowing of the physis and a high Ellipsoidal Index, which may be indicative of poor prognosis. Levels of Evidence for Primary Research Question: Level III, case-control study.