The corticosterone response to the sight of a natural predator was investigated in free-living and captive great tits (Parus major). Free-living great tits responded to the sight of a stuffed, slowly moving Tengmalm's owl, a major predator of great tits, with warning calls and a change in behaviour around a feeder. Great tits returned to the feeder within a few minutes and began to approach the owl, and there was no increase in plasma corticosterone levels in birds sampled 30-50 min after they first saw the owl. Captive great tits in an aviary were exposed for 30 min to a stuffed Tengmalm's owl, to a stuffed brambling, and to a cardboard box. All three stimulus objects were slowly rotated during the exposure period. Great tits exposed to the owl changed their behaviour immediately, and spent most of the time when the owl was visible flying around the aviary and hanging from the roof, with very few visits to a feeder. Great tits exposed to the brambling and to the moving box also changed their behaviour and made fewer visits to the feeder. The great tits responded to the sight of the owl with a marked increase in plasma corticosterone levels, whereas there was no change in corticosterone levels (mean levels < 11 ng/ml) in birds exposed to the brambling or to the moving box. Mean corticosterone levels were high (37.1 +/- 4.9 ng/ml) 0.5 h after exposure to the owl, remained high (38.9 +/- 6.0 ng/ml) 1 h after exposure, and had returned to basal (5.3 plus minus 1.3 ng/ml) 3 h after exposure to the owl. This is the first demonstration for any bird of a complete corticosterone response to a predator. The sight of a predator initiated a corticosterone response in great tits that could not move more than 3 m away, whereas free-living great tits that could choose how far to fly away from the predator either did not initiate a corticosterone response, or had a small corticosterone response in which corticosterone levels were not significantly different from basal 30-50 min later. The results indicate that the initiation of a corticosterone response in birds depends on whether or not a bird perceives that a stimulus is a threat. Furthermore, they illustrate the importance of not making generalised conclusions based on laboratory experiments.