Investigations into plant-pathogen interactions have provided us with several models underlying the genetic basis of host resistance in plants. In the past decade, tens of resistance genes have been isolated from numerous crop and model plant species and these form a few distinct classes when classified by domain structure, the majority being nucleotide-bindingsite- leucine-rich-repeat (NBS-LRR) genes. The NBS-LRR family consists of two sub-families based on the N-terminal domain: the coiled-coil (CC) NBS-LRRs and the Toll Interleukin Receptor homology domain (TIR) NBS-LRRs. The potential of these genes for future and current agricultural breeding programs has driven a large number of studies exploring the members of these gene families in the genomes of a variety of crop species. In the present study I focused on the NBS-LRR family in the allohexaploid wheat genome and obtained a comprehensive set of Triticeae NBS-LRR homologues using a combination of data-mining approaches. As starting point I detected conserved motifs in the dataset, finding all six previously characterized in the core-NBS domain of other plant NBS-LRRs. Phylogenetic analysis was performed to study relationships between the Triticeae NBS-LRR family and the 25 CC-NBS-LRR (CNL) R genes identified to date. I found the Triticeae CNL family to be highly divergent, containing ancient clade lineages, as seen in all angiosperm 120 taxa previously studied, and found a number of “ancient” dicotyl R genes grouped with Triticeae clades. The evolution of recent NBS-LRR gene duplications in the Triticeae was studied at the hand of two modes of duplication - firstly individual gene duplications yielding paralogous loci and secondly gene duplication by allopolyploidy. Current models of NBS-LRR family evolution predict that functional divergence occurs after gene duplication. An alternative is that divergence takes place at allele level, followed by a locus duplication that fixes heterozygosity in a single haplotype by unequal recombination. I investigated this hypothesis by studying the evolution of gene duplicates in two different contexts – paralogous duplications in the diploid barley genome and homeologous duplications in the allohexaploid genome of wheat. Nonsynonymous to synonymous substitution rate ratios were estimated for paralogous gene duplications in three recently diverged NBS-LRR clades. All pairwise comparisons yielded Ka:Ks ratios strongly indicative of purifying selection. Given that R gene mediated resistance is inherited qualitatively rather than quantitatively, I interpret this as evidence that even closely related paralogous copies (90-95% identity) should have independent recognition specificities maintained by purifying selection. Homeologous duplications were studied in allohexaploid wheat (AABBDD) using a section of the go35 NBS-LRR gene (2L) of the B and D diploid donor species of wheat. Numerous synonymous substitutions distinguished the B and D genome copies, with an absence of nonsynonymous substitutions. In contrast, single unique nonsynonymous substitutions were found in four out of five polyploid wheat go35 alleles, indicating that selection pressure was indeed relaxed across the homeolocus. Recent studies on polyploid genomes have shown that duplicated resistance genes are far more likely to be eliminated than highly transcribed genes such as tRNAs and rRNAs. These results are in agreement with the view that functional divergence takes place before duplication for NBS-LRR genes, as the loci duplicated by polyploidy appear not to evolve under purifying selection, as I found for the paralogous loci investigated.