Performance in a visual mental rotation (MR) task has been reported to predict the ability to recognize retrograde-transformed melodies. The current study investigated the effects of melodic structure on the MR of sequentially presented visual patterns. Each trial consisted of a five-segment sequentially presented visual pattern (standard) followed by a five-tone melody that was either identical in structure to the standard or its retrograde. A visual target pattern was either the rotated version of the standard or unrelated to it. The task was to indicate whether the target pattern was a rotated version of the standard or not. Periodic patterns were not rotated but melodies facilitated the rotation of non-periodic patterns. For these, rotation latency was determined by a quantitative index of complexity (number of runs). This study provides the first experimental confirmation for cross-modal facilitation of MR.