The regression of age at menarche on the difference between bone age and chronological age was estimated using hand/wrist radiographs taken between the ages of 10 and 15 years of 352 British girls whose ages at menarche were known. The regression was used to determine, in the case of girls who had not yet menstruated at a given chronological age and bone age, (a) the most probable age at menarche, and (b) the age by which 95 per cent would have experienced menarche. The results were compared with predictions made on the basis of chronological age alone. The range of error of predictions based on chronological age only decreases as age increases. This is true also when bone age is taken into account but, at all ages, the error of prediction using bone age as well as chronological age was less than that of prediction based on chronological age alone.