Cultivars of Miscanthus used as bioenergy crops or tested in trials are largely clonally propagated, wild sourced genotypes or clonally propagated F1 hybrids. One of the most productive taxa is the sterile triploid M. × giganteus. Little domestication or breeding has been undertaken and there is huge potential to utilize the extensive genetic resources of the genus for crop improvement. The challenge is to generate new highly adapted genotypes suitable for a range of environments. Production on marginal land, not used for food crops, is particularly desirable, but presents many barriers to crop breeders, as these are largely unproductive and/or stressful environments. This article outlines progress made in characterizing natural genetic variation in Miscanthus including next-generation single-nucleotide polymorphism genotyping, quantitative trait locus analysis and association mapping. It also explains how this knowledge is being used to develop novel genotypes suited for growth in a broad range of agricultural and marginal lands by defining breeding pools, generating novel crosses, manipulating polyploidy and applying genomic selection approaches.