Two methods for initiating epizootics ofEntomophthora muscae(Cohn)Fresenius in house fly,Musca domesticaL., populations were evaluated on New York dairy farms. First, 500 fresh, laboratory-infected fly cadavers were placed weekly for 10wk inside barns in areas of high fly activity. Second, 2,500 living, infected flies were released on each of two occasions spaced 3 wk apart. Releases were started in mid-June. Three farms were used for each treatment plus three control farms. TheE. muscaeused for the releases had been isolated from house flies collected from dairies the year before and had been maintained by continuous fly-to-fly passage. Prevalence rates on both types ofE. muscaerelease farms were twice as high (23–28%) as on control farms (12%)in the weeks after the start of the releases. Releases were more effective on farms with high fly populations than on farms with low fly populations. Examination of conidia indicated that the released strain (8–18 nuclei per conidium) became established as a result of the releases. Epizootics occurred on all farms, including controls, in September through November. Infections in the spring were exclusively with a strain ofE. muscaewith 2–8 nuclei per conidium; by fall, most infections (86–93%) were with a strain with 8–18 nuclei per conidium, even on control farms. Nuclear densities in the laboratory culture remained constant at 8–18 nuclei per conidium during this shift in wild infections. Neither of theE. muscaetreatments significantly reduced fly populations.