The widespread use of genomic tools has allowed for a deeper understanding of the genetics and the evolutionary dynamics of domestication. Recent studies have suggested that multiple domestications and introgression are more common than previously thought. However, the ability to correctly infer the domestication process depends on having an adequate representation of wild relatives. Cultivated maize (Zea mays spp. mays) is one of the most important crops in the world, with a long and a relatively well documented history of domestication. The current consensus points towards a single domestication event from teosinte Zea mays spp. parviglumis from the Balsas Basin in Southwest central Mexico. However, the underlying diversity of teosintes from lowlands Z. mays spp. parviglumis and highlands Zea mays spp. mexicana) was not taken into account in early studies. We used 32,739 SNPs obtained from 29 teosinte populations and 43 maize landraces to explore the relationship between wild and cultivated members of Zea. We then inferred levels of gene flow among teosinte populations and maize, the degree of population structure of Zea mays subspecies, and potential domestication location of maize. We confirmed a strong geographic structure within Z. mays spp. parviglumis and documented multiple gene flow events with other members of the genus, including an event between Z. mays spp. mexicana and maize. Our results suggest that the likely ancestor of maize may have been domesticated not in the Balsas Basin as previously thought, but in Jalisco or in the Pacific coast, and that different populations of the teosinte subspecies have contributed to modern maize’s gene pool. Our results points towards a long period of domestication marked by gene flow with wild relatives, making domestication a long and ongoing process.