Background. The distribution of birth weights has been extensively investigated. A common result of these studies is that the distributions differ from the Gaussian. A standard attempt has been to split the distribution into two components: a main distribution assumed to be normal and the other a small correction component. Studies have also shown an association between the distribution of birth weights and gestation age.

Methods and results. We consider data from the Åland Islands (Finland) for the period 1885-1998. The birth rate showed significant seasonality, having two peaks, one from March to May and the other from July to September, but only slightly significant seasonality in the birth weights was obtained. We observed significant correlations between birth weight and year of birth, sex of the newborn and type of maternity (single or multiple), condition of the infant (live or stillborn), maternal age and marital status of the mother. We built a multiple regression model and included time, maternal age, sex of the newborn and type of maternity as influential regressors. We also included regional grouping and found significant regional variations in birth weight. Marital status of the mother and condition of the infant at birth (live or stillborn) had too many missing values for inclusion in the regression model, but the effects of these factors have been analysed separately. The association between birth weight and gestation age was also explored.

Discussion. Comparisons with birth weights presented in data sets from other populations were also performed.

maternal age, sex, type of birth, seasonality, regional variation, temporal trends, regression analysis.