songs that are not notated but transmitted through live performance are of particular interest for the psychological study of the stability of tempo across multiple performances. While experimental research points to highly accurate memory for the tempi of well-known recorded music, this study asks whether there is any evidence of absolute tempo in a performance tradition that does not draw on such reference recordings. Fifty-four field recordings of performances of one Aboriginal dance-song, Djanba 14 , were analyzed. Results showed that over a span of 34 years, performance tempi deviated positively or negatively, on average, by 2%. Such small tempo variation is similar to JND thresholds to discriminate the tempi of isochronous sequences. Thirty-five field recordings of another song from the same repertory, Djanba 12 , deviated in tempi by an average of 3%. We discuss the musical, psychological, physical, and cultural factors likely to shape such temporal stability.