This paper reviews recent genetic evidence for the origins of the traditional cultivated bananas of the Pacific, and shows that they are unexpectedly complex. Current assumption of their prevailing west-to-east spread from Southeast Asia into the Pacific thus needs modification. Although bananas are widely assumed to have been part of the set of crops transported to Polynesia at first settlement, the linguistic evidence on which this is based underestimates the diversity of bananas in the New Guinea region and is suspect. Archaeological evidence of bananas is so far very tenuous. Recent genetic evidence of the parentage of most groups of cultivated bananas shows that the primary step toward edibility occurred in the Philippines– New Guinea region. Early movements westward across Island Southeast Asia must have occurred, and the complexity of hybrids makes regionally dispersed development likely. There is no demonstrable link with Taiwan or the adjacent coast of China. There is no evidence that the genetically distinct lineages of bananas found in Polynesia were brought together in the putatively ancestral Lapita crop assemblage of the northern New Guinea region. The complex phylogeny of the cultivated Pacific bananas may thus suggest multiple prehistoric introductions of bananas to Polynesia. If bananas were part of the founding set of crops of Remote Oceania, the question ''which bananas?'' is currently unanswered.