The objective of the research work was to estimate genetic effects and to analyze the factors associated with the genotype by environment interaction in a population by tester assessment in maize. Thirty-three maize populations were crossed with six populations used as testers, where parents and top crosses were set up in replicated experiments for agronomic traits evaluation during 2005. The experiments were established at three contrasting environments, two irrigated and one under rain fed conditions using an incomplete block design. Populations, testers, and testers x populations were statistically different (P ≤ 0.01), including the interaction with sites. The trait x environment biplot obtained by the rank correlation coefficients between the grain yield and the traits per environment allowed to identify similar testers as low yielding (CPRE, CINTER, and CAFIME), and high yielding (CTROP, 6221, and 6222). Population combining ability effects across environments by testers were different to those obtained by each of the low and high yielding group of testers, indicating that the two sets of testers discriminated the populations differently. Similarly, two heterotic groups of testers were defined based on the specific combining ability and heterosis among the testers by the populations evaluated, which allowed to identify sets of crosses according to their specific performance. Testers contributed differently to the relative magnitude of the environmental variance as a response to the environmental conditions, being the populations 6221 and 6222 the most sensitive.