ABSTRACT The search for productive and stable genotypes is the main goal of breeding programs. The Genotype × Environment interaction strongly influences genotype performance, and makes the selection of new cultivars difficult. One way to take advantage of this interaction is to identify genotypes with high grain yield (GY) and stability in different environments. The objective of this study was to evaluate the consistency of correlation between GY and stability evaluation methods in multi-environment trials and identify which methods are more suitable for selecting genotypes. GY data from 11 soybean (Glycine max [L.] Merr.) cultivars conducted in Value for Cultivation and Use trials in 10 locations in Paraná and Mato Grosso do Sul states, Brazil, in the 2013-2014, 2014-2015, and 2015-2016 crop seasons. A randomized complete block design with three replicates was used. Seven methods were applied to evaluate stability, and Spearman's correlation coefficient was used to compare methods. Positive associations were observed between GY and the harmonic mean of genotypic values (HMGV) across environments and genotype main effect + Genotype × Environment interaction effect by ideal genotype (GGE IG) methodologies, and between GY and the Lin and Binns method modified by Carneiro for general and unfavorable environments. The Eberhart and Russell, additive main effects and multiplicative interaction (AMMI1), and GGE for stability (GGE STA) methods presented no positive associations with GY in any year. Positive associations were found between the Wricke, AMMI1, and Eberhart and Russell methods because they were related to the static stability concept. The HMGV and GGE IG methods can be used together because they are consistently associated with GY and based on the dynamic stability concept.