Plant germplasm collections can be invaluable resources to plant breeders, provided they are well-characterized. After 140 years of acquisition and curation efforts by a wide and largely non-coordinated array of private and institutional actors, the current US collection of cold-hardy kiwifruit (Actinidia spp.) is rife with misclassifications, misnomers, and mix-ups. To facilitate the systematic improvement and resource-efficient curation of these species of long-recognized horticultural potential, we used genotyping-by-sequencing (GBS) data to deconvolute this historic collection. Evaluation of a total of 138 accessions (103 A. arguta, 28 A. kolomikta, and 7 A. polygama) with an interspecific set of 1,040 high-quality SNPs resulted in clear resolution of the three species. Intraspecific analysis (2,964 SNPs) within A. arguta revealed a significant level of redundancy (41.7%; only 60 unique genotypes out of 103 analyzed) and a sub-population structure reflecting likely geographic provenance, phenotypic classes, and hybrid pedigree. For A. kolomikta (3,425 SNPs), the level of accession redundancy was even higher (53.6%; 13 unique genotypes out of 28 analyzed); but no sub-structure was detected. Numerous instances were discovered of distinct genotypes sharing a common name, different names assigned to the same genotype, mistaken species assignments, and incorrect gender records, all critical information for both breeders and curators. In terms of method, this study demonstrates the practical and cost-effective use of GBS data to characterize plant genetic resources, despite ploidy differences and the lack of reference genomes. With the recent prohibition on further imports of Actinidia plant material into the country and with the active eradication of historic vines looming, this analysis of the US cold-hardy kiwifruit germplasm collection provides a timely assessment of the genetic resource base of an emerging, high-value specialty crop.