The enormous diversity of cassava cultivars used by farmers in Brazil is partially justified by the fact that these cultivars have specific adaptations for each region and hardly a cultivar will be have in the same way in the different ecosystems. This shows the versatility and importance of conserving the genetic variability of the species, and for this main tenance there are Active Banks of Germplasms and Didactic Collections in Universities, in addition to the farmers' own contribution. One of the tools that help to identify differences among individuals is the morphological characterization, which seeks to study the external and internal structures of the plant through the acquisition of qualitative or quantitative data of inheritable characteristics. Thus, the present work aimed to analyze the genetic variability among 22 accessions of cassava belonging to the Didactic Collection of Cassava Germplasm of UFERSA by means of 37 morphological descriptors. For this, the experiment was installed in the Didactic Horta of the Federal Rural University of the Semi-Arid (UFERSA) in continuous rows spaced between 1.00 m and 0.60 m between pits. For each access, 13 pits were planted using two pits per pit with five to seven buds each. The characterization was carried out in three stages and according to the development of the plant at three months (apical leaf color, pubescence of the apical bud) at six months (leaf retention; central lobe shape; petiole color; Foliar lobe length, phylloxia length, petiole length) and at nine months (stem cortex color, stem epidermis color, external stem color, habit of stem growth, color of terminal branches of adult plants, height of plant, height of first branch, habit of branching, number of tuberous roots, number of commercial roots, extension of peduncle of root, shape of root, external color of root, root pulp color, root cortex color, cortex bark ease, root cortex thickness, number of tuberous plants per pit, fresh mass of commercial and non-commercial roots, diameter and length of the cortex, HCN content in root pulp and leaf; Brix and acidity). The hierarchical groupings of the accessions were obtained by UPGMA methods. The joint analysis of the quantitative and qualitative data for the determination of the genetic distance was based on the Gower algorithm. The relative contribution of the quantitative descriptors to the genetic divergence between the accessions was determined. We verified the formation of four groups for the quantitative variables and five for the qualitative variables. The joint analysis of the qualitative and quantitative data to determine the genetic distance, based on the algorithm of Gower, generated five groups. The descriptor that contributed the most to the genetic diversity among the accessions was the fresh mass of commercial roots (88.46%). In view of the results, the existence of genetic divergence among the accessions of cassava from the Didactic Collection of UFERSA was verified