Aim : To examine factors affecting birthweight of Aboriginal infants in the Kimberley region of north-west Australia. Research design : A retrospective study of maternal and infant health records obtained through routine data collection. Subjects and methods : Birthweight and length of 2959 infants born to 1822 women from 1986 to 1994 were analysed. Mothers and infants were matched using unique identification codes. Index births and subsequent births to the same mother were collated in order to examine relative birth order effects and to calculate birth intervals. Results : Regression analysis showed significant associations between weight of the index birth and maternal age ( p < 0.001), remoteness of locality ( p < 0.01), sex of the infant ( p < 0.001) and maternal height ( p < 0.001). Length at birth was significantly associated with ethnicity of infant (Aboriginal vs Aboriginal/non-Aboriginal admixture, p < 0.05), sex ( p < 0.001), remoteness ( p < 0.01) and maternal height ( p < 0.001). Similar associations were observed for second and subsequent births. Birth interval was not associated with birthweight or length. Low birthweight was also more common to Aboriginal mothers compared with mothers of Aboriginal/non-Aboriginal admixture (13.1% vs 9.2%; h 2 = 5.1, p < 0.025) even though there were no differences in height between these two groups. Teenage mothers ( r 19 years) were no more likely to have low birthweight babies than older mothers. Of the variables examined, the only significant predictor of low birthweight was a previous low birthweight baby (relative risk = 4.45, p < 0.001). Conclusions : Short birth intervals and teenage births were not significant contributors to low birthweight in the present study. The high prevalence and duration of breastfeeding in the Kimberley may contribute to long average birth intervals. Pre-term birth, rather than intrauterine growth retardation, is likely to be the most common cause of low birthweight in this population.