The genotype-environment interaction is considered, by most geneticists and plant breeders, as one of the limiting factors to improve genotypes. Among the various methods for interpreting and analyzing genotype-environment interaction, the approach most often used seems to be the multiplicative model. In this paper a recent version known as the AMMI model, in which principal components are used, is applied to analyze the lint yield performance of eight cotton genotypes (six developed in Cd. Obregon, Son. Mexico and two developed in Casa Grande, Arizona by Deltapine Co.), in nine environments across northern Mexico,. The first principal component, which accounted for 45% of the genotype-environment interaction sum of squares, has its first six elements (corresponding to the Yaqui Valley environments) positives and negatives the corresponding to other environments; this indicates that there is a significant different response of genotypes in the Yaqui Valley and the other environments. The most stable genotypes were intermediate or full season specially adapted for the Yaqui Valley, but some early genotypes despite being less stable have very good yield potential. The highest yielding cultivars were those developed in the Yaqui Valley, outperforming Deltapine types. Early genotypes seem to perform better in environments different from the Yaqui valley.