BACKGROUND: Detection of abnormal fetal growth is vital to antenatal care, and traditionally birthweights that are <10th or >90th centile are classified as small or large for gestational age (LGA). Evidence regarding outcomes for birthweight centiles outside these extremes remains unclear. AIMS: To evaluate the relationship between birthweight centile and perinatal death and determine the 'optimum' birthweight centile with the lowest rate of perinatal mortality. METHOD: Data on all Victorian births from 1999 to 2008 were stratified into smaller subsets than the traditional small for gestational age (SGA) (<10th centile), appropriate for gestational age (AGA) (10-90th centile) and LGA (>90th centile) and analysed by all gestations, for term births alone, and using the 'fetus at risk' approach. Multiple logistic regression was used to adjust for age, parity and co-morbidities. RESULTS: For term births, the 'optimum' birthweight centile was the 50-90th range (1.1 perinatal deaths/1000 births). Lower birthweight centiles had significantly higher rates of perinatal death - even those that would be classified as AGA. Babies with a 10-25th birthweight centile had a two-fold increased risk of perinatal death (AOR 2.10, 95% CI 1.6, 2.7). Even those with a 25-50th birthweight centile had higher perinatal mortality rates (AOR 1.58, 95% CI 1.3, 2.0). There was no strong evidence of higher perinatal mortality in larger birthweight centiles, except term births >99th centile. The 'fetus at risk' analysis showed a rise in perinatal mortality after 37 weeks' gestation for all birthweight centiles, particularly for SGA babies. CONCLUSION: Babies with a birthweight below the 50th centile are at greater risk of perinatal mortality compared with the 'optimum' ≥50 to <90th centile group.