DescriptionFungi in the genus Colletotrichum are destructive pathogens, afflicting over 900 plant species worldwide, including numerous grasses and cereal crops. In this dissertation I reconstruct the evolution of grass-inhabiting Colletotrichum and relate it to pathogenicity. I have undertaken this research to (1) understand how graminicolous Colletotrichum were shaped by evolutionary processes such as population divergence, gene flow, mutation, recombination and speciation; and (2) evaluate whether the lifestyles of these fungi can be correlated with genotypic or genomic signatures, life history or ecological adaptations. These broad objectives were centered on the application-based model of C. cereale, a recently emerged pathogen of the golf course turfgrasses. Novel tools, including sequence-based markers from four protein coding genes and five transposon species, transposon RFLPs, and microsatellite markers were developed and deployed. Although ITS sequences are currently the most common method of classifying Colletrichum species, this research highlighted the potential inaccuracy of ITS-based classification. The unreliability of 47% of Colletotrichum ITS sequences from public databases suggested a proliferation of compromised species identifications. Comparison with multilocus phylogenies showed Colletotrichum ITS data are insufficient for the task of taxonomic resolution. In addition, these studies demonstrated traditional classification tools (i.e., morphology and host range) are subject to convergent evolution. Phylogenetic reconstructions showed that graminicolous Colletotrichum underwent a prominent historical split, separating cool-season grass-associated taxa from lineages inhabiting warm-season grasses. Eight novel Colletotrichum species were identified and described; two species were emended. C. cereale populations were found in native grasses, cereal crops and turfgrass environments, but disease was limited to turfgrass. Genotypic data, along with the detection of the meiosis-specific repeat-induced point mutation process provided evidence of recombination in C. cereale, a fungus long presumed asexual. Extreme differentiation between locally-adapted populations indicated that asymptomatic grasses are unlikely reservoirs of infectious disease propagules that could initiate disease in turf. But gene flow from the generalist C. cereale founder population and specialized genotypes provides an indirect pathway for genetic exchange between otherwise isolated populations. Together, these studies contribute substantially to the growing number of genomic resources available for this increasingly important evolutionary research system.