In this article I argue that pre-transitional natural fertility was sometimes controlled through birth-spacing, and that coitus interruptus was probably an important means of such control. First, the motivation for spacing was often strong, although not necessarily associated with a desire for family limitation; and control through spacing, although much harder to detect than a parity-dependent deceleration of the rate of childbearing, has been identified unequivocally in some pre-transitional and transitional populations. Secondly, coitus interruptus is reasonably effective and harmonizes with ancient and persistent notions of reproductive physiology; and evidence for its use comes both from statements of criticism and advocacy, and from a rich set of metaphors and euphemisms. The scenario of natural fertility controlled by means such as coitus interruptus is not offered as an alternative to the emergence of family limitation, which was probably the great innovation of the fertility transition. Rather, it directs us to recognize the diversity of the childbearing behaviour of the past, just as we recognize that of today.