The genotype by environment interaction (GEI) represents a main difficulty in the selection of promising genotypes for different environments. The objective of this research was to evaluate the GEI and the productive potential of 25 starch maize varieties, using the additive main effects and multiplicative interaction (AMMI) model. The information was collected from five trials in different locations of the Tayacaja province, Peru, during the 2018-2019 season, under an alfa lattice 5x5 experimental design with three replications and experimental units of two rows 4 m long, 0.80 m between rows and 0.20 m between seeding points. The cultivar productive potential was measured through the grain yield, adjusted to 15 % moisture. Once the GEI was defined, the multivariate analysis was carried out to obtain the singular values of the first significant model terms for genotypes and environments. The GEI was highly significant and explained around 14 % of the yield phenotypic variation, discriminating some varieties with specific adaptation, while others showed broad adaptation to the environments. The AMMI model explained about 96 % of the variation due to the GEI, showing that only the first two axes concentrated about 85 % of that variation. The local varieties Chullpi-Q, Carhuay-P, Astilla Blanca-Cusqueado, Astilla Blanca-DH, Cusqueado-P and Astilla Blanca-Astilla Amarilla showed the best productive potential, but only the first three remained stable through the evaluation environments.