Fertility exposure analysis was developed recently by Hobcraft and Little in order to evaluate the relative importance of the various proximate determinants upon levels of current fertility. In the present paper we extend the analysis to the study of changes in fertility between two cross-sectional surveys. We show how to express fertility change as a product of terms which represent changes in the proximate determinants. The model of change is adapted to a log-linear framework, in which the proximate determinants on the one hand, and socio-economic variables such as education, on the other, are considered together. A few simplifications of the method are also suggested so that fewer demands are made on data. The model and its possible interpretations are illustrated with pairs of surveys from Pakistan and Mexico. In each pair, one survey was part of the WFS programme and the other was very similar but conducted five years later.