A lateral cephalometric radiograph is frequently performed for diagnostic and pretherapeutic purposes in patients with obstructive sleep apnea syndrome (OSAS). We studied the prognostic value of this exploration in terms of therapeutic outcome in surgically treated patients. Fifty-five patients underwent surgery from May 1994 through December 1998. Forty-seven had phase I surgery (UPPP, hyothyrohyoidopexy and genioglossal advancement), 18 phase II surgery (bimaxillar advancement) after failure of a phase I procedure and 8 primary phase II surgery. For the "phase I" group: mean body mass index (BMI) was 26.3+/-2.9 kg/m2 and mean age was 47+/-11 years. For the "phase II" group: mean BMI was 25.9+/-3 kg/m2 and mean age was 48+/-9 years. Polysomnography was performed in all patients preoperatively and six months after each surgical procedure. The preoperative apnea-hypopnea index (AHI) was 45.2+/-26.8/h of sleep for the phase I group and 53.8+/-26.9/h for the phase II group. All the patients had a lateral cephalometric radiograph, preoperatively, postoperatively, and at 6 months. The following parameters were measured on each radiograph: posterior airway space (PAS), mandibular plane-hyoid bone distance (MPH), minimal retrolingual space, minimal retrovelar space, surfaces of the rhinopharynx, the oropharynx, the hypopharynx and total upper airway surface. Therapeutic success was defined as a AHI<15/h and 50% decrease compared with the preoperative AHI, associated with normal sleep structure, respiratory microarousal score less than 15/h, normal oxymetry and absence of symptoms. After phase I surgery: the success rate was 21.2%. For the whole group, the total upper airway surface has significatively increased between preoperative and immediate postoperative time, as well the MPH, the PAS and the minimal retrolingual space. But at the late postoperative control, no significative difference compared with the preoperative data has been observed. The comparison between failures and successes has demonstrated that there was no difference in surface or distance benefit between the two groups. But it existed a preoperative difference as the failures have a greater rhinopharynx and a shorter retro velar oropharynx compared with the successes. This difference has been noticed in the immediate postoperative time but not in the late postoperative time. In the "failure" group, the immediate postoperative increase in the upper airway surface, the PAS and the minimal retrolingual space was totally lost in the late postoperative control. In the opposite, in the "success" group, the minimal retro velar space was the only parameter significatively increased at the postoperative time. After phase II surgery: the success rate was 76.9%. All measured parameters except rhinopharynx surface and MPH were increased at last follow-up; part of the increase in the hypopharynx and the minimal retrolingual space observed postoperatively was lost during later follow-up. Nevertheless, in the "failure" group patients, no significant increase could be demonstrated at the last postoperative control. Linear parameters (PAS, minimal retro lingual and retro velar spaces) were smaller in the "successful" group than in the "failure" group. It is difficult to ascertain the exact contribution of the lateral cephalometric radiograph to the assessment of surgical outcome. Apparently, and independently of the technique used, part of the gain in the upper airway surface observed immediately after surgery is progressively lost. We were unable to define any parameter on the lateral cephalometric radiograph predictive of success after phase I surgery. Discrimination between success and failure after phase I surgery might be related to the stability of the increase in the minimal retro velar space and the MPH. For phase II surgery, the initial shortness of the upper airway surface is a good prognostic factor for therapeutic success, defined as a stable increase in the oropharynx.