Gene flow between maize [Zea mays (L.)] and its wild relatives does occur, but at very low frequencies. Experiments were undertaken in Tapachula, Nayarit, Mexico to investigate gene flow between a hybrid maize, landraces of maize and teosinte (Z. mays ssp. mexicana, races Chalco and Central Plateau). Hybridization, flowering synchrony, pollen size and longevity, silk elongation rates, silk and trichome lengths and tassel diameter and morphology were measured. Hybrid and open-pollinated maize ears produced a mean of 8 and 11 seeds per ear, respectively, when hand-pollinated with teosinte pollen, which is approximately 1-2% of the ovules normally produced on a hybrid maize ear. Teosinte ears produced a mean of 0.2-0.3 seeds per ear when pollinated with maize pollen, which is more than one-fold fewer seeds than produced on a maize ear pollinated with teosinte pollen. The pollination rate on a per plant basis was similar in the context of a maize plant with 400-500 seeds and a teosinte plant with 30-40 inflorescences and 9-12 fruitcases per inflorescence. A number of other factors also influenced gene-flow direction: (1) between 90% and 95% of the fruitcases produced on teosinte that was fertilized by maize pollen were sterile; (2) teosinte collections were made in an area where incompatibility systems that limit fertilization are present; (3) silk longevity was much shorter for teosinte than for maize (approx. 4 days vs. approx. 11 days); (4) teosinte produced more pollen on a per plant basis than the landraces and commercial hybrid maize; (5) teosinte frequently produced lateral branches with silks close to a terminal tassel producing pollen. Collectively these factors tend to favor crossing in the direction of teosinte to maize. Our results support the hypothesis that gene flow and the subsequent introgression of maize genes into teosinte populations most probably results from crosses where teosinte first pollinates maize. The resultant hybrids then backcross with teosinte to introgress the maize genes into the teosinte genome. This approach would slow introgression and may help explain why teosinte continues to co-exist as a separate entity even though it normally grows in the vicinity of much larger populations of maize.