This study compares bone mass in a national sample of girls with Rett syndrome (RS) with a sample of control children. The Australian RS Database was the source of cases for this population-based study. Hand radiographs were available from 101 of 137 subjects (74% of the known Australian population of girls with RS aged < or = 20 years). Control radiographs matched for age, sex, and laterality were obtained from hospital radiology departments. A measure of cortical thickness was made from the difference between the outer diameter and the medullary space in the second metacarpal bone. A mean z-score value for cortical thickness and percentage cortical area for each individual was calculated. The mean cortical thickness (z score) for girls with RS was -1.94 compared with -0.38 for control children (P<0.001). In girls with RS, the mean cortical thickness decreased with age (P<0.001). In girls who were taking epilepsy medication it was -2.21 compared with -1.23 in those not taking epilepsy medication (P<0.001). There was no evidence of a beneficial effect of increased calcium intake on cortical thickness. A similar pattern was obtained when percentage cortical area was estimated. In multivariate analysis, increasing age and use of anticonvulsant medication were associated with decreased cortical thickness and only use of anticonvulsant medication with decreased percentage cortical area. Fractures had occurred in one-third of cases and it was estimated that just over 40% of girls would sustain a fracture by the age of 15 years. Girls with RS may be at increased risk of fractures and their bone quality compromised as determined by cortical thickness and percentage cortical area measurements from the second metacarpal.