Creating specialized lines is one of the techniques of genetic improvement and conservation of breeds and populations of the honeybee. The aim of this study was a comparative assessment of the diversity and differentiation degree of A. m. caucasica lines based on morphometric analysis and microsatellites (MS). Material for this study was the worker bees of Gray Mountain Caucasian breed (lines I-V, n = 728) which were selected in five apiaries in the Greater Sochi of Krasnodar krai. Morphometric analysis included such measurements as the length of the proboscis (LP, mm), width of third tergite (W3T, mm) and cubital index (CI). Molecular genetic studies were based on seven MS loci (A024, A88, A113, AP043, HB-C16-05, HB-THE-03, HB-C16-01). The level of variation between families for morphometric parameters was determined by two-way hierarchical analysis of variance. Genetic differences between families for MS were estimated by paired comparison of Fst values. Fst matrix was used for PCA-analysis. To determine the quantitative estimation of variation between families within lines we calculated Fst, Rst (AMOVA). The degree of line’s differentiation for morphometric characters was evaluated by calculating the Euclidean distances. The obtained values were used to construct a dendrogram of similarity by a single bond (single linkage) of the hierarchical clustering algorithm. The differentiation of lines for MS was based on calculating the values of Nei genetic distances. Similarity dendrogram was constructed using the method of UP- GMA. We performed summary statistic using the software STATISTICA, GenAlEx (v. 6.5.1), PAST (v. 3.03). Morphometric analysis showed the presence of significant differences between the lines for LP and W3T whereas there was no difference between the lines for CI. The greater heterogeneity concerning studied traits in the lines II and V was revealed, and on the contrary, there was more consolidation in the lines III and IV. Bees of line I differed significantly from the rest of the lines on both traits, but they were characterized by significant differences between families in LP. Analysis of MS profiles showed similar trends in assessing the level of intra- and interfamily variability. We observed an excess of heterozygotes in the line I (Fis = 0.048), which can be considered as an indication of the high heterogeneity. Bees of this line were characterized by a minimal individual (Fit = 0.052) and the maximal interfamily variability (Fst = 0,124). Lines II-V were characterized by a deficiency of heterozygotes (Fis = 0.062-0.128), a relatively higher individual variability (Fit = 0.143- 0.189) and lower values of interfamily variability comparing to line I (Fst = 0.095-0.104). The lowest interfamily differences were observed in the lines III and IV (Fst = 0.096 and 0.095, respectively). Analysis of the differentiation of the studied lines for morphometric characteristics and MS revealed differences in the structure of the family tree. The dendrogram based on MS data is a reflection of the geographical origin of these lines. The structure of the family tree, based on morphometric characters, does not reflect the geographic closeness (differences) of origin or similarities (differences) in the economically useful traits of studied lines. Thus, the results of our studies of the morphometric parameters and MS show similar trends in assessing intra- and interline variation, but there are differences in assessing differentiation of lines using two methods. In the future complex approach will allow to identify not only breeds of bees with high accuracy, but also smaller taxonomic units. It is hoped that the research results in general can be used in breeding work to restore the purity of the honeybee breeds.