A study was made of the seasonal variation in all births, and births according to marital status, multiplicity and birth status (live and still) in Switzerland recorded between 1876 and 1990. To obtain seasonal variation in as pure as possible form, our analyses are based on rates. When comparing the seasonality in data sets showing markedly different levels, standardised indices were used. Assuming the length of pregnancies with twins to be about one month shorter than for pregnancies with singletons, lagged twinning rates were calculated but, in comparison with actual twinning rates, the general seasonal variation remained. Therefore, this study was based on actual twinning rates. A monotonic increase in the amplitude of the seasonal variation in general births was noted for the period 1876-1930, with strong seasonal variation holding for 1921-1980. After that, a marked decline in the amplitude can be observed. Seasonality of both all births and twin maternities showed very similar pattern for the periods 1876-1930 and 1969-1990, with maxima in the spring (March-May) and troughs in late autumn (October-December). Twin maternities showed a strong seasonality for the period 1876-1930, being about 20% higher in March than in October. The twinning rate in the period 1876-1930 was about 2.6 per thousand units higher than in the period 1969-90. For twin maternities there was also a stronger seasonal variation during the earlier period than during the later one. The pattern of the seasonal variation for extramarital births, showing a maximum in February (conceptions in May-June) and a minimum in August (conceptions in November-December) with a difference of no less than 24% was more marked than for the marital births. It seems likely that this seasonality of extra-marital maternities was due mainly to seasonal variation of coital rates and multiple ovulation in the early summer months coinciding with optima of light, temperature and food supply. A strong reduction in the rate of stillbirths (gestational age more than 29 weeks) was observed during the twentieth century. The stillbirth rate declined from about 40 per 1000 in the 1870s to fewer than 5 per 1000 in the 1980s. Irrespective of this strong decline in the stillbirth rate, the same seasonal rhythm was noticed throughout the period with high stillbirth rates among births around March and low rates during the summer and autumn.