Background: Small babies are known to be at increased risk for a wide range of difficulties. In predicting risk, it may be more informative to estimate smallness in relation to family norm (using the birth weight of an older sibling) rather than to use the standard "small-for-gestational-age" (SGA z-score) measure. Methods: For 10,577 babies born to women enrolled in the Danish National Birth Cohort, we calculated a "birth-weight ratio" (actual birth weight/expected birth weight predicted from older sibling X 100). We identified babies in the lowest decile of the birth-weight ratio (≤87.7% for boys and ≤87.6% for girls) and compared them with babies in the lowest decile of the sex- and gestational age-specific distribution (SGA z-score). We evaluated how these definitions predicted motor development. We also compared how selected predictors of birth weight influenced the classification of the baby according to the birth-weight ratio and according to the SGA z-score. Results: Birth-weight ratio and SGA identified 1058 and 1059 babies, respectively, with 738 identified by both methods. A low birth-weight ratio predicted delayed motor development slightly better than did SGA. Babies classified as too small solely by the SGA criterion were more often born to small or smoking mothers than those identified solely by the birth-weight ratio. Conclusions: The combined use of the birth-weight ratio and SGA may provide a more sensitive tool for identifying babies at risk. The birth-weight ratio is less likely to classify babies as small based on their mothers' body size, identifying instead babies who were substantially smaller than their sibling.