Producing language involves the real-time sequencing of words into phrases, leading to considerable demands on working memory that can be relieved by ordering biases in spoken utterances. One such bias is called easy-first: the tendency for more easily-accessible phrases to occur earlier in an utterance, allowing for incremental planning of more complex phrases (MacDonald, 2013). Recent evidence suggests that this bias may extend beyond language to effect other domains involving real-time action sequencing. In the current study, we sought to test for the presence of the easy-first bias in a creative domain that similarly requires real-time action sequencing: musical improvisation (e.g. Pressing, 1987). Using a corpus of 456 transcribed improvisations from eminent jazz musicians (e.g., Charlie Parker, John Coltrane), we tested for easy-first on multiple definitions of easiness over the phrase and over the corpus: interval frequency, interval size, interval variety, pitch variety, and direction changes. Similar to language production, our findings suggest that expert improvisers consistently retrieve “easier” melodic sequences before generating more complex and novel sequences, indicating a similarity in the domain-general sequencing biases that facilitate the spontaneous production of music and language.