As early as infancy, humans are sensitive to prosodic cues that can aid recognition of words embedded in sentence contexts. We addressed the question of whether similar cues formed by expressive nuances in music performances aid listeners in recognition of musical phrases embedded in melodic contexts. Utilizing a task similar to infant-research habituation paradigms, we report experiments with musically experienced and inexperienced listeners who are familiarized with performances of musical phrases that were identical in pitch/duration contents but differed in their intensity and articulation cues. Listeners then completed a recognition task for performances of the same musical excerpt whose cues either matched or did not match the performances at familiarization, and whose cues were either consistent or inconsistent with the rhythmic context in which the excerpts were embedded. Findings show that listeners can distinguish musical phrases that differ only in expressive nuances, and a mismatch of expressive nuances to the rhythmic context can facilitate recognition. These findings suggest that the prosodic cues that differentiate human performances are part of listeners’ memory for melodies, and similar acoustic features may enable the recognition of auditory events in music as in speech. [Work supported by NIMH.]