Identification of quantitative trait loci is one of the important issues in breeding program especially under stress condition. Plant material in the present study was a natural population including 72 tobacco genotypes. In order to apply osmotic stress, Mannitol with 0.33 Molar concentrations was used. The genotyping of individual were performed using 21 out of 35 primer combinations of AFLP markers. After measuring of eight traits including total dry weight, dry weight of shoot, dry weight of leaves, length of highest root, total dry weight of root, length of longest leaf, length of longest stem and dry weight of root to dry weight of shoot ratio, in order to perform association analysis and identify linked markers controlling loci five statistical models were used by TASSEL software. These models were GLM and MLM using data set of phenotype and genotype data, population structure data (Q), principle components (PC) and Kinship data derived from general similarity in genetic background arising from shared kinship (K). For all eight traits common in all models and the markers with explanation of the most phenotype variations were identified and discussed. For plant total dry weight E110-M160-3, for root dry weight E080-M150-4 and for dry weight of shoot E080-M140-23 with the highest coefficient of determination (19.9, 23 and 14.9% respectively) were identified as the most important markers. In total, the numbers of identified markers by GLM were more than MLM model. Considering the results, it will be expected that detected AFLP markers in the present study can be used in marker assisted selection programs after confirmation and specification being specified with high confidence.