The hypothesis that anencephalus stems from fetus-fetus interactions in dizygotic twin pairs is examined by comparing the epidemiological predictions of the hypothesis with available observations. The hypothesis itself was based upon the disease-discordance and sex-concordance characteristics of twin pairs affected with anencephalus, and upon the sex ratio of the disease itself. The testable predictions of the hypothesis are (a) that variations in the incidence of anencephalus should be related to variations in dizygotic twinning rates, and particularly that dizygotic twinning rates will set upper limits to the incidence of anencephalus, and (b) that the F/M ratio in anencephalic infants will be high in circumstances of high incidence, and specifically when the incidence is high in relation to the dizygotic twinning rate. Examples from international comparisons, secular changes, social class gradients, and variations according to maternal age confirm a consistent correspondence between observations and these predictions. In addition, the possibility was tested that some fetus-fetus interactions might be based upon sequential rather than simultaneous pairs of fetuses. This model predicted asymmetries of sex ratio in sibs immediately preceding propositi, with differences according to the sex of the affected child, and the predicted findings were confirmed. The fetus-fetus interaction hypothesis is therefore extended in the terms that about one third of occurrences are determined in a sequential manner. The success of the extended fetus-fetus interaction model in explaining a large number of otherwise unrelatable findings confirms its validity.