There is a general consensus in the literature that fertility differences between populations can be accounted for by differences in just four key proximate determinants: nuptiality, the postpartum non-susceptible period, contraception and abortion. Natural fecundibility is generally assumed to be constant between populations. This paper puts the theoretical and empirical case for a re-evaluation of that assumption, drawing on the under-utilized data on sexual activity collected in the Demographic Health Surveys (DHSs). Using data for married women in nine African countries, the analysis finds substantial population level differences in mean monthly coital frequency, which, if accurate, suggest an important demographic effect. There is a clear regional patterning to these differences, with levels of activity considerably lower among women in the West African populations included in the study than those from East and southern Africa. For West Africa in particular the data indicate the normality of exceptionally long periods of very infrequent or no intercourse by married women outside the period of postpartum abstinence. The findings challenge prevailing presumptions concerning susceptibility to pregnancy in marriage on which statistics for unmet need for family planning are derived. While doubts are raised over the precision of the sexual activity data used, the paper argues for the need for a greater effort to operationalize the 'proximate determinant of conception', not only for more accurate fertility modelling, but also as a planning tool for a more sensitive provision of family planning services in Africa.