Ornamental grasses may become invasive weeds depending on their ability to naturalize and outcompete other species. Miscanthus sacchariflorus (Maxim) Franch. (Amur silvergrass) is a tall, self-incompatible, nonnative grass that has become naturalized in eastern North America, forming monospecific stands and raising concerns about its potential invasiveness. To understand the extent of clonal and sexual reproduction in feral populations, we examined their population genetic structure, seed production, and ploidy. We surveyed 21 populations in Iowa and Minnesota using eight polymorphic microsatellite markers. Only 34 multilocus genotypes (MLGs) were detected from a total of 390 samples. All of the study populations had more than one MLG, thereby allowing cross-pollination with near neighbors, but most were dominated by one or a few MLGs. Low genetic divergence suggests that all populations may have originated from similar cultivars. Cluster analysis showed that the six populations from Minnesota were extremely similar to each other, whereas the 15 populations from Iowa were somewhat more diverse. Seed production was quantified for 20 populations and ploidy for 11 populations. Average seed production was very low (< 0.30 seeds per panicle), although most populations did produce seeds. Because the populations were diploid (2x), they also may have the potential to hybridize with ornamental varieties of Miscanthus sinensis (Chinese silvergrass; eulaliagrass), a diploid close relative. Clonal growth, self-incompatibility, and spatial isolation of compatible clones may contribute to pollen-limited seed set in these populations. Low seed set may affect the rate of spread of feral M. sacchariflorus, which appears to disperse vegetatively as well as by seed. Although this species is not widely viewed as invasive, it is worth monitoring as a species that may become more widespread in the future.