Craniofacial and anthropometric characteristics are identified risk factors for obstructive sleep apnea syndrome (OSAS). Phase 1 of this study used cephalometric radiographs to record craniofacial measurements from 60 OSAS subjects with a respiratory disturbances index (RDI) of 20 or greater (group A) and 60 subjects with a history of loud snoring and an RDI less than 20 (group B). From this data set, a craniofacial risk index (CRI) was constructed using age, body mass index (BMI), and 14 cephalometric measures previously reported to be associated with OSAS. A separating boundary (CRIS) was established by using discriminant analysis to differentiate between the two groups. All measurements were determined by an investigator who was blinded to the subjects' RDI score. Phase 2 used a second sample of 19 group A and 47 group B subjects to test the ability of the CRI derived from the first sample to classify subjects in this second sample into the correct age group. The CRI was able to classify correctly 72.3% (34 of 47) of the group A subjects when all variables were used in the discriminant model. Using only four variables (age, BMI, hyoid mandibular plane distance, and tongue length) selected by the stepwise method, 72.3% (34 of 47) of the group B subjects and 78.7% of group A subjects were classified correctly. These results suggest that a stepwise CRI could be used to classify heterogeneous groups of individuals with increased RDI into subgroups with varying degrees of anatomic risk for disease. Such subgrouping by anatomic risk could be important in determining the pathophysiology of OSAS because it is likely that differences in upper airway anatomy among individuals interacts with a variety of other factors to produce clinical illness.