In this research 45 single crosses from a diallel of ten subtropical maize lines from CIMMYT and three checks were evaluated in five sites. The experimental design utilized was an Alfa lattice with two replications per location. Combined analysis across locations were made with a randomized complete block design. Method IV of Griffing’s was used to calculate estimates of general (GCA) and specific combining ability (SCA). The best crosses were 5x10, 5x8, 5x6, 4x10 and 5x7 with a yield of 11 t/ha aprox. The best cross was 5 x 10 which was 21% superior in seed yield and with better agronomic characteristics than the check CML78 x CML97. The line S89500F2-2-2-2-2-B-B showed the highest GCA effect. Two subtropical heterotic groups were formed with the following lines: “A”, lines one, four, five, and six; group “B” , lines two, three, seven, eight, nine, and ten. Line five can be used as a tester “A” and the line ten can be used as a tester “B”. The hybrids and lines selected could be used in subtropical environments.