The average national sweet potato yield of Brazil falls below the productive potential of the crop because of the cultivation of local and unimproved varieties. To improve this, more productive cultivars must be adopted along with adequate culture treatments. This study was conducted between January and May 2019 in Selvíria, Mato Grosso do Sul, Brazil, to characterize sweet potato genotypes obtained through crossbreeding. The experimental design consisted of randomized blocks containing 264 genotypes, the control (‘Beauregard’), and two replicates. Plant harvesting began 127 d after planting. After harvesting, the roots were washed and dried in a covered area ready for evaluation. The total, commercial, and non-commercial yield; total, commercial, and non-commercial root number; root dry matter content; and dry matter productivity were evaluated. The genotypes CERAT16-20, CERAT31-1, and CERAT21-2 are promising in terms of root production for household consumption because of their high productivity of commercial roots. In contrast, genotypes CERAT16-20, CERAT31-1, CERAT25-17, CERAT25-12, CERAT21-2, CERAT29-26, CERAT34- 4, CERAT31-11, and CERAT24-8 are promising for industry because of the high production of dry mass per hectare. The main components, total number of commercial roots, production of non-commercial roots, mass of commercial roots, total production of dry mass of roots, mass of roots, and total production of roots have a low contribution to the discrimination of the genotypes; therefore, their analysis can be discarded in future studies, under the same soil and climate conditions, thus reducing workload, expense, and time.