Although originally domesticated in tropical America, the sweetpotato [Ipomoea batatas (L.) Lam.] has a long history of cultivation in the Pacific region. While the post-Columbus dispersal of sweetpotato to Asia and the Pacific is well documented, the hypothesis that there was a prehistoric transfer of sweetpotato by Peruvian or Polynesian voyagers from Peru to Oceania has long been a controversial issue. The objective of this study was to assess the genetic diversity and interrelationships of sweetpotato landraces from the Pacific region and Latin America, and test the hypothesis of human transfer of this crop to the Pacific Island in prehistoric times. Seventy-five sweetpotato landraces from Peru, Ecuador, Mexico, the Philippines, Papua New Guinea (PNG) and 5 Oceania countries were analyzed using amplified fragment length polymorphism (AFLP). Multidimensional scaling (MDS) and analysis of molecular variance (AMOVA) revealed a large genetic variation in the Oceania gene pool, far greater than that in Peru-Ecuador. The Mexican cultivars were grouped together with those of Oceania. In contrast, there is little association between the Peru-Ecuador germplasm and that of Oceania. These results suggest that Peru-Ecuador may not be the source of the Oceania sweetpotato germplasm. Natural dispersal from Mesoamerica is an alternative explanation, to the 'Kumara hypothesis', for the origin of the Oceania sweetpotato.