1. Costs of reproduction have been assessed experimentally by measuring subsequent survival and reproduction of parent animals raising enlarged and reduced numbers of offspring. Reported effects on survival have so far always referred to local survival of marked individuals in the study population. They do not provide definitive proof of a cost of reproduction, since reduced local survival may be due either to reduced survival or to an increased tendency to emigrate from the study area. Therefore, it is important to assess mortality rates in connection with brood size experiments. 2. We report an analysis of the time of death in 63 cases where kestrels, Falco tinnunculus L. had raised broods of manipulated size and were subsequently reported freshly dead. 60% of the parents raising two extra nestlings were reported dead before the end of the first winter, compared to 29% of those raising control or reduced broods. This result confirms our interpretation of the manipulation effects on local survival as due to mortality rather than emigration. The extra mortality occurred in the winter following the brood enlargement. 3. Kestrel parents in these experiments have been shown to adjust their daily energy expenditure to the modified brood size. Increased parental effort in this species thus entails an increased risk of death half a year later