ABSTRACT. In this study, we aimed to establish strategies for value for cultivation and use (VCU) experiments for the tobacco crop in the southern region of Brazil with respect to the number of environments used to assess tobacco lines. Trials of the Virginia (18 sites) and Burley (17 sites) varietal groups were conducted in the three states of the southern region of Brazil in the 2009-2010 crop season. The experiment was conducted in a completely randomized block design with four replications of 10 tobacco lines in the final stage of evaluation; the plots had 6 rows of 7 plants each, or 42 plants per plot. The cured leaf weight per hectare (kg/ha) was obtained. To evaluate stability, the ecovalence and additive main effects and multiplicative interaction models were adopted. In addition, joint analyses of variance were carried out considering different site numbers by simulating resampling. The site number ranged from 2 to 17 or 2 to 16, depending on the varietal group, and sites were selected at random without replacement. The process was repeated 2000 times for each number of sites. All analyses were performed using the R software. The results are very similar for both varietal groups. There is no advantage of using a large number of sites for VCU experiments in the southern region of Brazil because many sites contributed little to the interaction or did not discriminate the tobacco lines. Furthermore, the classification of the best lines is very similar to that obtained in the total number of evaluated sites.