Development of core collections is an effective tool to extensively characterize large germplasm collections, and the use of a mini-core subsampling strategy further increases the effectiveness of genetic diversity analysis at detailed phenotype and molecular levels. We report the formation of a mini-core subset containing 217 entries derived from 1794 core entries representing the genetic diversity found in more than 18,000 accessions of the USDA-ARS rice (Oryza sativa L.) germplasm collection. The mini-core was developed with PowerCore software on the basis of 26 phenotypic traits and 70 molecular markers. The 217 entries in the mini-core had a similar distribution over 15 geographical regions, with 1794 entries in the original core collection. The resultant mini-core had 6.3% of mean difference (MD%), 16.5% of variance difference (VD%), 102.7% of variable rate (VR%), and 97.5% of coincidence rate (CR%) with the core collection, which brought about full coverage of 26 traits. All 962 alleles identified by the 70 markers in the core collection were captured in the mini-core, which maximized allelic richness up to 100% and resulted in an average genetic diversity (Nei index) of 0.76, ranging from 0.37 to 0.97 among the markers. In conclusion, the mini-core presented in this study is a highly suitable and representative subset of the USDA rice core collection as well as the entire USDA-ARS rice germplasm holdings.