This paper investigates the role of lexical ambiguity in the processing and recognition of multimodal advertisements. First, we combined 28 Russian advertising posters: 14 ads with an ambiguous headline that leads to the conflict between text and pictural parts of a poster and 14 explicit ads of the same structure and advertising product as their ambiguous pairs. In Exp.1, 104 Russian participants looked at the stimuli and rated the stimuli by 6 parameters: how eye-catching, original, impelling to buy the advertised product, understandable, evoking positive emotions and matching the product the posters are. We used scales from 1 to 5. The experiment was conducted via Google Forms. 3–4 weeks later, in Exp. 2, the same participants were invited to recognize verbal and non-verbal “halves” of the posters from Exp. 1. The results of the scaling experiment showed that ambiguous advertising posters were ranked significantly higher than unambiguous ads by four of six parameters. Furthermore, ambiguous posters were rated as slightly more understandable than unambiguous posters, despite the expectation that word-play would complicate the understanding of the ambiguous posters. In the recognition task, we revealed that both verbal and non-verbal components of a poster are recognized better if they were a part of an ambiguous poster. These findings are discussed in relation to the multimodal text literature, and research perspectives are provided to explore ambiguous text processing empirically.