Several traits must be considered simultaneously in selection of tobacco cultivars for meeting the requirements of farmers, industries and consumers. In this context, it is important to obtain information to improve the efficiency of the tobacco breeding programs. Experiments with tobacco hybrids are usually developed in several locations and years, a strategy that aims to increase selection efficiency, considering the effects of genotype x environment interaction. The complexity of identifying superior genotypes is simplified with the use of methodologies based on Mixed Linear Models, which allow modeling fixed and random effects simultaneously. In this research, the objective was to evaluate the performance of 24 Burley tobacco hybrids, generated by the Alliance One International, regarding the yield, nicotine and leaf quality Index characters, in seven environments in the southern region of Brazil, using a Mixed Linear Model approach. The tests were carried out in a randomized complete blocks design, with three replications, together with five commercial controls. Statistical analyses were performed in the JMP environment, using Mixed Linear Models. Based on the models selected by character, the average predicted values of the hybrids were obtained in the average of the environments and within each environment, making it possible to select the most adapted hybrids for each location. The use of the statistical approach of Mixed Linear Models proved to be effective with regard to the selection of superior genotypes, given the nature of the data set, which had imbalance in terms of the evaluated genotypes, in addition to a large number of environments where the tests were conducted. There were significant differences for the effect of genotype for the evaluated characters, which indicates that for these characters there is genetic variability. The behavior of the hybrids within the environments was verified, allowing the selection of hybrids according to the studied environments.