College education has a positive impact on birth rates, net of age and duration since previous birth, according to models estimated separately for second and third births. There are also indications of such effects on first-birth rates, in the upper 20s and 30s. Whereas a high fertility among the better educated perhaps could be explained by socioeconomic or ideational factors, it might just as well be a result of selection. When all three parity transitions are modelled jointly, with a common unobserved factor included, negative effects of educational level appear. On the whole, the effects are less clearly negative for women born in the 1950s than for those born in the 1940s or late 1930s. The cohorts from the 1950s show educational differentials in completed fertility that are quite small and to a large extent stem from a higher proportion of childlessness among the better-educated. Second-birth progression ratios are just as high for the college educated as for women with only compulsory education, and the third-birth progression ratios differ very little. This reflects weakly negative net effects of education after first birth and spill-over effects from the higher age at first birth, counterbalanced by differential selectivity of earlier parity transitions.