Objective To investigate the clinical characteristics and outcomes of pediatric Langerhans cell histiocytosis (LCH) with craniofacial bone involvement. Methods A retrospective analysis was performed on 145 pediatric LCH patients with craniofacial bone involvement registered at Beijing Children′s Hospital Affiliated to Capital Medical University from January 2007 to July 2013.The patients were divided into 2 groups: central nervous system risk craniofacial bone involvement group(CNS-RISK) and non-central nervous system risk craniofacial bone involvement group(non-CNS-RISK). All patients were assessed at 5 weeks, 11 weeks, 25 weeks and 52 weeks respectively after chemotherapy started, and 3 months, 6 months, 1 year and 3 years after chemotherapy withdrawal.Statistics and related risk analysis was performed respectively. Results A total of 145 craniofacial bone involved LCH cases were included, which was composed of 62.5% of 232 LCH cases hospitalized during the same period.The median age of these patients was 29 months, and median follow-up time period was 31 months.The most commonly involved craniofacial bone was parietal bone(78 cases, 53.8%), followed by temporal bone(59 cases, 40.7%) and frontal bone(57 cases, 39.3%). The onset age was significantly different (26 months vs.54 months, Z=-2.777, P<0.05) between CNS-RISK group (103 cases) and non-CNS-RISK group (42 cases). Moreover, compared with non-CNS-RISK group, CNS-RISK group showed higher ratio of patients classified as multisystem involvement of risk organs (72/103 cases, 69.9%)vs.(15/42 cases, 35.7%)(χ2=16.908, P<0.05), and a higher rate of overall relapse rate (45/103 cases, 43.7%) vs. (7/42 cases, 16.7%) (χ2=9.427, P<0.05), a lower survival rate of 3-year relapse-free survival rate [(66.9±5.7)% vs.(88.2±7.8)%, Z=2.205, P<0.05]. The incidence of diabetes insipidus was 13.7% in 232 LCH patients.Compared with patients without craniofacial bone involvement, patients with craniofacial bone involvement demonstrated a higher rate of diabetes insipidus [(27/145 cases, 18.6%) vs.(5/87 cases, 5.7%), χ2=7.579, P=0.006]. But the incidence of diabetes insipidus showed no statistical difference between CNS-RISK group and non-CNS-RISK group (21.3% and 11.9%, χ2=1.760, P=0.185). Diabetes insipidus was not found in single system LCH with Single-Bone CNS-RISK lesions.Till the end of follow-up, 1 out of 145 patients died.Among 145 patients, 5 cases had a single-bone CNS-RISK lesion.They received systemic chemotherapy.One showed reactivation, and none of them died.Multivariate analysis of variance showed that all the independent factors indicating diabetes insipidus included parietal bone, frontal bone, maxilla and mandible involvement(HR=2.697, 3.487, 5.425, all P<0.05), while independent factors indicating relapse included temporal bone, maxilla and mandible involvement(HR=3.712, 3.380, all P<0.05). Conclusions Among involved craniofacial bones, the parietal bone is most commonly involved.LCH occurs averagely at an earlier age in CNS-RISK group, along with lower 3-year relapse-free survival rate, high relapse rate, and more patients classified as multisystem LCH involvement of risk organs.The incidence of diabetes insipidus in children with craniofacial bone involvement with single system CNS-RISK is low.Patients with the parietal bone, frontal bone, maxilla and mandible involvement at diagnosis are at a increasing risk a significantly to develop DI during the course of disease. Key words: Langerhans cell histiocytosis; Child; Craniofacial bone; Clinical characteristics.