Verbal overshadowing refers to the surprising effect whereby additional verbal information about a visual stimulus hinders its subsequent recognition. In two experiments, we analyzed the validity of this effect for event recognition across various conditions of presentation and testing. Participants observed events that were either followed (Experiment 1) or preceded (Experiment 2) by a verbal description. Results showed that verbal overshadowing occurred when the verbal description was presented after the visual presentation, independent of the distractor type. However, when the verbal description preceded the event, recognition performance was seen to improve when distractor items incompatible with the verbal description were used. The findings were interpreted in terms of two interacting mental representations, which differ both in their level of abstraction and in their accessibility.