In order to test the hypothesis that the trichothecene genotype composition of local populations of Fusarium graminearum is structured by specific habitats, a collection of 1,407 isolates was obtained from overwintered maize stubble, mature maize ears and wheat spikes, and the atmosphere 1.5 m aboveground during the flowering stage of these crops. These isolates were sampled at three diverse agricultural locations in New York State: namely, Aurora (sampled in 2012 and 2013) in central New York, Belmont (sampled in 2013) in southwestern New York, and Willsboro (sampled in 2013) in northeastern New York. Approximately 100 isolates of F. graminearum from each habitat were collected within a 10-mile2 area in each location. Polymerase chain reaction assays were used to identify three main B-trichothecene genotypes--3-acetyldeoxynivalenol (3-ADON), 15-ADON, or nivalenol (NIV)--based on amplification of portions of Tri3 and Tri12 genes. All but the NIV genotype were detected. The 15-ADON genotype predominated in most locations; frequencies were 92% (652/709) at Aurora, 78% (332/379) at Belmont, and 53% (167/319) at Willsboro. Frequencies of any genotype did not differ in general among the four habits in each location. An exception was in Aurora 2012, where only 5 in 24 3-ADON isolates were found in samplings from the air and grains of both crops. As viewed by the composition of trichothecene genotypes, local populations of F. graminearum appear not to be structured by these four habitats inclusive of pathogenic and saprophytic phases of the fungus life cycle. The similar frequency of 3-ADON and 15-ADON in eastern New York (Willsboro), which is less than 400 km away from the Aurora sampling location in the central area of the state, suggests that regional populations may be differentiated based on selection associated with climatic or landscape features not currently identified.