We studied whether the presence of breeding kestrels (Falco tinnunculus) affected nest predation and breeding habitat selection of curlews (Numenius arquata) on an open flat farmland area in western Finland. We searched for nests of curlews from an area of 6 km2 during 1985-1993. For each nest found, we recorded the fate of the nest, and the distance to the nearest kestrel nest and to the nearest perch. We measured the impact of breeding kestrels on nest predation by constructing artificial curlew nests in the vicinity of ten kestrel nests in 1993. Curlew nests were closer to kestrel nests than expected from random distribution, eventhough kestrels fed on average 5.5% of curlew chick production. Predation risk by kestrels was lower than predation risk by corvids and other generalist predators, which predated 9% of curlew nests surviving farming practices and an unknown proportion of chicks. Artificial nest experiment showed that nest predation was lower close to kestrel nests than further away suggesting that the breeding association of curlews and kestrels was a behavioural adaptation against nest predation. Thus, the presence of a predator may sometimes be beneficial to prey, and prey animals have behavioural adaptations to these situations.