Mobbing is a widespread anti-predator strategy in birds, and predators are generally expected to avoid mobbing. For example, observational studies suggest that the cryptic roosting behaviour of nocturnal predators, such as many owls, may be a strategy to limit mobbing. In this paper, we present the results of the first experimental study investigating to what degree roost exposure influences the risk of being mobbed, and the intensity of a mobbing incidence once initiated. To determine these factors, we used an experimental setup with taxidermic mounts of tawny owls Strix aluco in Grib Skov forest, Denmark. The risk of an owl being mobbed during a 50 min morning survey period increased with the exposure of its roosting position, from 24% when hidden to 85% when openly exposed. The corresponding increase in the afternoon was from 6% to 36%. This suggests that an owl may minimize the mobbing rate by reducing the encounter rate with potential mobbers through its choice of roost. Once initiated, the duration of the mobbing (a proxy for the presumed cost of being mobbed) was independent of the roosting position of the mounted owl, but was positively correlated with the number of birds in the mob.