Summary Many recent fertility studies in developing societies put forward the hypothesis of a negative relation between economic class and fertility. Data showing a positive relationship are frequently dismissed a priori as resulting from the reporting errors of illiterate women. This study draws on data from Indonesia's 1971 Census, a 1973 sample survey of fertility and mortality, and an intensive community study in Java, to argue that an observed positive relation between class and fertility is real, and is related to differences in patterns of marital disruption, postpartum abstinence, and fecundity. The positive relation may be reversed in the future as changes in these patterns, and the impact of the national family planning programme, affect the family structure of each class differently. Had the positive relation in this context been attributed offhand to reporting errors, these important socio-economic changes would have been misunderstood, and possibly ignored.