ABSTRACT The agronomic traits have complex quantitative inheritance, wheat breeding has a highly cost and time expending to achieve advanced lines, and new cultivars. The objective of this work was to estimate the phenotypic (rp), genotypic (rg) and environmental (re) correlation coefficients and the genetic parameters, as well as to estimate the selection gains between agronomic traits in wheat using multivariate analysis. F6 generation was carried out in a randomized block experimental design, with fifteen wheat genotypes arranged in three replications. The traits measured were days of emergence to flowering, plant height, number of tillers, spike length, number of spikelets, kernel weight and thousand kernel weight. Correlation coefficients, variance components, genetic parameters, selection gain, Euclidiana’s distance, relative contribution of traits and canonical variables were estimated. Selection based on number of tillers, kernel weight and thousand kernel weight may result in expressive selection gains. The number of fertile tillers per plant showed a significant and positive intermediate association with kernel weight and positive correlation with number of spikelets. The cycle showed greater contribution to the genetic divergence among genotypes studied. Distinct groups highlighted the genetic variability among genotypes.