A computer model developed to simulate the interaction between the pathogenic fungus, Entomophaga maimaiga, and the gypsy moth, Lymantria dispar, was used to investigate airborne dispersal of fungal conidial spores. The model used data on egg mass density gathered from 32 sites in an area of Connecticut and information on fungal prevalence, gypsy moth abundance, fungus resting spore load in the soil, and detailed weather records from six intensively sampled plots in the same area. It calculated seasonal survival rates of gypsy moths at the six plots using a variety of dispersal distributions, or kernels, for the conidia. Distributions ranged from normal densities to the exponential, a Bessel and power distribution, along with a “kinked” linear distribution, which had two parts. All kernels gave good fits to the data so long as the dispersion parameter caused each standardized distribution to have an abscissa value near 0.05 at a distance of 1.25 km from the source. However, significant dispersal beyond 10 km only occurred for the kinked linear and power distributions. Thus, mechanisms for short-range dispersal may be different from those for long-distance dispersal. The model was also used to investigate the potential for dispersal of E. maimaiga in the northeastern United States just after it was known to be established in 1989. For the weather conditions prevalent in 1989 and 1990, and assuming a kinked linear dispersal kernel, I predicted that the fungus would spread rapidly, which did indeed happen. Furthermore, in these years rainfall and other weather conditions were very favorable for fungus development, so even if relatively few conidia dispersed long distances, they might easily have initiated viable infections.