A preselected cluster of births (12351-4719 = 7632 = n) which was divided into 3 groups (delivery between 28-32 gestational weeks, 33-37 gestational weeks and > 37 gestational weeks) has been correlated with six major weather situations of the four meteorologically defined seasons. The correlation was made with the date of birth and with one day as well as two days before. The duration of pregnancy was longer by an average of 0.45 weeks (i.e. 3 days) when the major weather situation did not change for more than 8 days. The group with delivery 28-32 weeks is more sensitive to meteorological influences than the group with delivery 33-37 weeks. During cyclonic as well as during anticyclonic atmospheric drifts the groups 28-33 wks and 33-37 wks have been statistically over-represented. As far as the season is concerned over-representation is found in autumn. We conclude that pregnancies at risk of premature delivery should be followed up in short intervals during cyclonic and anticyclonic atmospheric drifts especially during the autumn season.