Understanding the genetic diversity and population structure can effectively improve the breeding efficiency of grain sorghum new cultivar. In this study, genotyping by sequencing (GBS) technology was deployed to unlock the genetic diversity in a collection of 120 grain sorghum materials. Within 3456 SNP markers that were generated, the polymorphism information content (PIC) ranged from 0.013 to 0.574, with the average of 0.381, while SNP-based genetic distance value ranged from 0.137 to 0.538, with the average of 0.353. Both phylogenetic tree analysis and principal component analysis indicated that 120 grain sorghum genotypes were clustered into three groups. Group-1 was mainly composed of sorghum materials with a relatively distant relationship, including the USA material MN-3609. Within Group-2 and group-3, these genotypes were mainly collected from northern and south parts of China, respectively. Population structure analysis showed that the maximum ΔK was observed for K=3, suggested three sub-populations detected in 120 grain sorghum materials. This result was consistent with phylogenetic tree analysis and principal component analysis. Taken together, this resulted deciphered the genetic background and population structure of grain sorghum from the genotype diversity level, which will provide theoretical basis in breeding for grain-use sorghum new cultivar in China.