The effect of maternal build on the outcome of pregnancy was studied in two birth cohorts in Northern Finland, for 1966 and 1985-86. Prospectively collected data were available for 10,969 women in the earlier cohort and 9128 in the later one. The women in the earlier cohort were on average 2.9 cm shorter but 0.2 kg thinner and had 0.7 kg/m2 greater BMI. 13% of the women in the earlier cohort had a BMI below 20, but 24% in the later one, while 96% in both cohorts had BMI below 30. The women with low BMI were on average taller than the others, and at all BMI levels the women of the earlier cohort were shorter and lighter than those of the later one. The outcome of pregnancy was measured by the incidence of pre-term births and perinatal plus childhood deaths up to the age of 4 years, and the association of maternal body measurements with low birth weight (< 2500 g) and small for gestational age (SGA) infants was also studied. An additive logistic regression model was fitted in each analysis, to determine the probability of the outcome separately in terms of BMI, weight and height, adjusting for maternal age, parity, smoking, marital status, father's social class and place of residence. No evidence was found that BMI values 20-25, commonly judged as optimum for the mother's own longevity, predicted a better prognosis for the child than values below 20.(ABSTRACT TRUNCATED AT 250 WORDS)