ABSTRACT This paper explores the suggestion (Steen, 2023a, 2023b) that most metaphor may be structurally ambiguous between deliberate and non-deliberate meanings, which in turn affords multivalent metaphor use. The paper begins by examining a sample of 56 Metaphor-Related Words in 25 examples of language use from corpus research about metaphor in discourse about cancer and the end of life (Semino et al., 2018). These data are analyzed by means of a new method proposed by Deliberate Metaphor Theory (Steen, 2023a; cf.; Reijnierse et al. 2019). Results show that all metaphors in the sample that are both polysemous and conventional can be given two interpretations, one non-deliberate and one deliberate. This ambiguity is largely corroborated by additional analysis of these same data by Wmatrix. Subsequent inclusion of all other cases from the same chapter in Semino et al. (2018) offers additional support. This finding offers a critical perspective on the idea that metaphor is a strong psychological device for figurative framing in discourse, for the structural ambiguity of most metaphor allows for multivalent metaphor use, where, in fact, most metaphor may typically be comprehended non-metaphorically (Steen, 2008, 2023a, 2023b). The ambiguity of metaphor hence also explains the paradox of metaphor.