The prevalence of breathing obstruction was determined in a cohort of 4-year-old children. Craniofacial morphology was studied in obstructed children and compared with data from a control group of 4-year-old children with ideal occlusion. Dental arch morphology was compared in obstructed and non-obstructed children in the group. Parents of 95.5 per cent of the study base of 644 children answered a questionnaire concerning their child's nocturnal behaviour and related questions. The 48 children who, based on parental report, snored every night or stopped breathing when snoring (the 'snoring group'), showed a higher rate of disturbed sleep, mouth-breathing, and a history of throat infections as compared with the rest of the cohort. These children were examined by both an orthodontist and an otorhinolaryngologist and, when indicated, they were also monitored in a sleep laboratory. Twenty-eight of the children were diagnosed as having a breathing obstruction (4.3 per cent of the cohort) and six children (0.9 per cent) had sleep apnoea (mean apnoea-hypopnoea index of 17.3), using the same definition as that for adults. Cephalometric values among the obstructed children differed from those of a Swedish sample of the same age with ideal occlusion. Thy had a smaller cranial base angle and a lower ratio of posterior/anterior total face height. Small, but not significant differences were seen for NSL-ML and NL-ML. Compared with 48 asymptomatic children from the same cohort, the obstructed children had a narrower maxilla, a deeper palatal height, and a shorter lower dental arch. In addition, the prevalence of lateral crossbite was significantly higher among the obstructed children.