With 2 figures and 2 tables The majority of functionally characterized plant disease resistance genes are of the nucleotide-binding site (NBS)-LRR gene family, encoding proteins with a central NBS domain, a carboxy-terminal leucine-rich repeat domain, and a variable N-terminal region with or without homology to the Toll interleukin 1-like receptor (TIR) domain, referred to as TIR and non-TIR resistance gene homologues (RGH), respectively. Degenerate primers designed from conserved motifs within the NBS sequence were used to amplify, clone and sequence NBS-RGH from the sweetpotato genome. Two hundred and twenty-five distinct sweetpotato NBS sequences with similarity to known RGH genes were identified. Additional 50 sweetpotato RGHs were mined from the public genomic sequence database. Thus, a total of 275 RGH sequences were obtained using both PCR-based method and data-mining approach, from which 237 were non-TIR sequences organized into 35 singletons and 35 groups after reduced to 90% nucleotide identity, and 38 were TIR sequences divided into three primary phylogenetic clades. A bias of non-TIR vs. TIR was observed not only in genomic RGH sequences, but also in expressed sequence tags-RGH sequences. A subset of sweetpotato non-TIR RGH genes contained a conserved intron within the NBS sequences. The exploration of RGH diversity enables resistance gene evolutionary study and may facilitate the isolation of new and functional alleles. These new RGH sequences provided a resource of candidate genes and molecular markers for disease resistance research in sweetpotato. © 2009 Blackwell Verlag GmbH.