We conducted a case control study of factors that may be related to th
e prevalence of multiple births with at least one liveborn. Over a per
iod of 15 months during 1991-1992 in Athens, Greece, we identified 380
women who had a multiple birth that was delivered in one of the six l
arger maternity hospitals in the area. We selected as controls 380 liv
ebirth singleton pregnancies delivered at the same hospitals immediate
ly following a multiple pregnancy. All mothers were interviewed in the
maternity wards or at home, and zygosity was ascertained through the
similarity method. The odds ratio for ovulation-inducing drugs was not
elevated for monozygotic twins, but it was 9-fold for dizygotic twins
and 90-fold for triplets or quadruplets, We excluded drug-induced pre
gnancies, twins of indeterminable zygosity, and all triplets from furt
her analyses. We found no remarkable association between any of the st
udy variables and the occurrence of monozygotic twins, with the except
ion of an increasing trend with age. For dizygotic twins, we found wea
k associations in the expected directions with maternal age (positive)
and height (positive) as well as with parity and number of induced ab
ortions (positive). A history of oral contraceptive use was inversely,
but weakly, associated with dizygotic twinning. Each cup of coffee pe
r day was associated with an increment in the odds ratio of 1.23-fold
(95% confidence interval = 1.05-1.43). Restriction of the analysis to
dizygotic twins of different gender increased the point estimate of th
e odds ratio to 1.31 for each cup of coffee consumed daily. There was
no association of tobacco smoking with dizygotic twinning, after contr
olling for coffee intake.