Article dans une revue: It is often assumed that early life circumstances, in particular before age two, are important for later human capital development. Using experimental variation in the timing of benefits from a conditional cash transfer program, we test the hypothesis that intervention starting in utero and continuing in the first two years is critical. At age ten, boys exposed to the program during this period had better cognitive, but not anthropometric, outcomes than those exposed in their second year of life or later. The lack of a differential effect on anthropometrics was due catch-up growth.