Jazz singing can be defined by the instrumental approach to the voice, where the singer can match the instruments in their stylistic approach to the lyrics, improvised or otherwise, or through scat singing; that is, the use of nonsensical meaningless non-morphemic syllables to imitate the sound of instruments.

The "roots" of jazz music were very much vocal, with "field hollers" and ceremonial chants, but whilst the blues maintained a strong vocal tradition, with singers such as Ma Rainey and Bessie Smith heavily influencing the progress of American popular music in general, early jazz bands only featured vocalists periodically, albeit those with a more bluesy tone of voice; one of the first Jazz recordings, the 1917 Original Dixieland Jass Band recordings featured one Sarah Martin as vocalist. It was Louis Armstrong who established singing as a distinct art form in jazz, realising that a singer could improvise in the same manner as instrumentalist, and establishing scat singing as a central pillar of the jazz vocal art. A frequently repeated legend alleges that Louis Armstrong invented scat singing when he dropped the lyric sheet whilst singing on his 1926 recording of "Heebie Jeebies". This story is false and Armstrong himself made no such claim. Jazz musicians Don Redman, Cliff Edwards, and Red Nichols all recorded examples of scat earlier than Armstrong. However, the record Heebie Jeebies and subsequent Armstrong recordings introduced scat singing to a wider audience and did much to popularize the style. Armstrong was an innovative singer who whilst experimenting with all kinds of sound, improvised with his voice as he did on his instrument. In one famous example, Armstrong scatted a passage on I'm A Ding Dong Daddy From Dumas he sings "I've done forgot the words!" in the middle of recording before taking off in scat.

Throughout most of jazz history, most vocal music was performed by either a soloist or a very small group of singers, usually one to a part. In the mid 1960s, a few brave directors decided to try transcribing Big band charts for voices. This idea was first tried by Hal Malcom at Mt. Hood Community College in 1967 with a group called Genesis. The group continues to host the oldest vocal jazz festival in the United States, the Northwest Vocal Jazz Festival. Hal Malcom directed the group for 23 years until passing it to his former student, Dave Barduhn, who continues to direct the group to this day. 13 years later, in 1980, Lonnie Cline formed a sister group at Clackamas Community College eventually taking the name Mainstream, which it performs under today. Mainstream was a new type of ensemble, one that moved away from performing the big band arrangements and other jazz standards and began to learn and perform a wide variety of music, including jazz, rock, pop, funk, and fusion. The idea was looked down upon at first but eventually was adopted by many and is now popular in high schools across the country. A vocal jazz ensemble usually consists of the choir, normally made up of 8-16 singers, and a Rhythm section most often consisting of a pianist, a bassist, a guitarist, and a drummer. Singers either each hold a microphone or sing with area microphones. Vocal Jazz ensembles also often sing a cappella.