The aim of this retrospective study was to test the reproducibility of three-dimensional (3D) mandibular cephalometric analysis in patients with retromicrognathia (Pierre Robin syndrome) compared with normal group of paediatric patients in the pre-orthodontic period of life. We analyzed 9 PRS patients (mean age: 7.75 years-old) who underwent cleft palate surgery in the first four months of life, performed by the same surgeon using the same technique. Cone-beam computed tomography (CBCT) was performed in these patients after local ethical committee approval. The normal group consisted of 15 patients (mean age: 10.16 years-old) with CBCT already performed for other reasons. Three-dimensional cephalometric mandibular analysis was performed with Maxilim software and consisted of 15 non-constructed landmarks identified directly on 3D CBCT reconstructions, 10 constructed landmarks belonging to planes, 4 planes, and 23 distances measurements. First, we tested the reproducibility of 15 non-constructed landmarks. Two independent observers performed the identification of 15 landmarks on 24 patients twice for each patient, with a one- week interval between the two series of measurements. Next, we tested the reproducibility of the 3D cephalometric mandibular analysis. Two independent observers performed the 3D cephalometric analysis for 24 patients twice for each patient, with a one- week interval between the two series of measurements. Concerning the reproducibility of the 15 non-constructed landmarks, the method used was at least as good in the PRS group as in the control group. PRS condition did not affect the difficulty of identifying and positioning the landmarks on 3D CBCT skull reconstructions. Concerning the reproducibility of 3D mandibular cephalometric analysis there was a significant difference in inter-observer measurements for 4 of 23 distance measurements ( p < 0.05) involving “gonion” and “condyle” landmarks. Reproducible measurement for the posterior vertical height of mandibular ramus should be based on other reference landmarks than “gonion” and condyle” landmarks.