Six experiments examined different determinants of metaphor performance. Special focus was placed on tenor‐vehicle similarities and on visual‐verbal strategy operations. The results showed that use of specified strategy operations tended to favour verbal operations, whereas previous research demonstrated that unspecified visual strategies improve metaphor comprehension. The deviating observations were interpreted as due to differences in levels of strategy specification. In the new experiments metaphors were shown to be less easily transformed under visual than under verbal processing conditions. The number of detected similarities between tenor and vehicle terms are shown to be only modestly related to metaphor performance. Findings suggested that similarities are more easily found with visually encoded metaphors, whereas an alternative tenor term is easier to find with the metaphor verbally encoded.