Water yam (Dioscorea alata L.) is among the most cultivated species used as a source of food and income for small-scale farmers in Tanzania. However, little is documented about Dioscorea species available in Tanzania, including their genetic diversity. This study used ten polymorphic microsatellite markers to determine the genetic diversity and relationship of 63 D. alata accessions from six major producing regions. Results revealed a polymorphic information content (PIC) of 0.63, while the number of alleles per locus ranged from 4 to 12 with a mean of 7.60. The expected heterozygosity ranged from 0.17 to 0.74, with a mean of 0.49, which suggests moderate genetic diversity of D. alata accessions. Kagera region had the highest mean number of (1.5) private alleles. Analysis of molecular variance revealed that 91% of the variation was attributed to within-population while among-population contributed 9% of the total variation. The highest Nei's genetic distance (0.65) was for accessions sampled from Arusha and Mtwara regions. Principal coordinate analysis and cluster analysis using Unweighted Paired Group Method using Arithmetic (UPGMA) grouped D. alata accessions into two major clusters regardless of geographical origin and local names. The Bayesian structure analysis confirmed the two clusters obtained in UPGMA and revealed an admixture of D. alata accessions in all six regions suggesting farmers' extensive exchange of planting materials. These results are helpful in the selection of D. alata accessions for breeding programs in Tanzania.