BACKGROUND Obstructive sleep apnea (OSA) is a common disorder characterized by recurrent episodes of a complete or partial collapse of the upper airway during sleep. Traditionally, the disease is diagnosed by overnight polysomnography. Studies have shown correlation between parameters of cephalometry and severity of sleep apnea. We wish to determine the variable of craniofacial dimensions in the upper airway that contribute to OSA, and to investigate the significance of craniofacial measurements in positional and non-positional sleep apnea patients. METHODS From July 2002 to June 2006, we studied 84 males and 15 females who came to the sleep center because of daytime sleepiness. All the participants underwent overnight polysomnography and lateral cephalograms, performed by an experienced technician. RESULTS Craniofacial measurements of gnathion-gonion, anterior superior hyoid to mandibular plane (MP-H), posterior nasal spine (PNS) to the velum tip (SPL), widest point of the soft palate (SPW), and the product of PNS to the velum tip and widest point of the soft palate (product of soft palate (SPP)=SPL x SPW) were positively related to the apnea/hypopnea index (AHI). The velum tip to the pharyngeal wall parallel to the Frankfurt horizontal (PAS) was negatively related to the AHI. We further divided the study subjects into 4 groups according to AHI value (group 1, AHI<5; group 2, 5 <or= AHI<15; group 3, 15 <or= AHI<30; group 4, AHI >or=30). Age, body mass index (BMI), neck circumference (NC), distances of PAS, SPL, SPW, SPP and angle of sella-nasion-infradentale (SNB) were significantly different depending on the degree of severity of sleep-disordered breathing (SDB). Patients who were older, with a high BMI and longer MP-H distance, had more daytime sleepiness (Epworth sleepiness scale, ESS). Furthermore, lower AHI values and longer PAS measurements were found in the positional sleep apnea group when compared to the non-positional sleep apnea group. After adjusting for confounding factors of age, BMI and NC, we found that BMI, MP-H distance and PAS measurement were correlated with severity of OSA. CONCLUSIONS Cephalometry could be a useful and inexpensive clinical tool to evaluate Chinese patients with OSA. MP-H and PAS should be measured in Chinese patients with OSA. MP-H was correlated with ESS. The PAS measurement was narrower in non-positional OSA patients compared to positional OSA patients.