Differences were found in habitat structure between nest sites of Emperor Geese (Chen canagicus) and Cackling Canada Geese (Branta canadensis minima) during a 5-year study (1982-1986) at Kokechik Bay, Alaska. Emperor Geese tended to select sites that afforded concealment from avian predators before incubation began. Cackling Canada Geese tended to select sites that enhanced avoidance of mammalian predators. Emperor Geese selected sites in areas away from open water, with tall dead vegetation adjacent to the nest site. Cackling Canada Geese selected sites close to shores of smaller ponds having more islands, and sometimes nested on islands, evidently enhancing avoidance of mammalian predators. During springs with heavy snow cover (>50% of surface covered on 25 May), both species selected sites appreciably different from sites selected during springs with light snow cover (<25% on 18 May). During heavy snow years, both species selected sites with more short, dead vegetation. Emperor Geese differed from Cackling Canada Geese by selecting sites with more shrubs and were farther from ponds. Cackling Canada Geese tended to nest closer together than Emperor Geese. Differences in structural habitat characteristics at nest sites of Cackling Canada Geese were not related to nesting success, but successful Emperor Geese selected nest sites with more and taller dead vegetation than unsuccessful pairs. Although it was possible to identify "typical" nest sites of Emperor Geese and Cackling Canada Geese, many sites were used by both species. Factors such as the amount and duration of snow cover, drainage patterns, presence or absence of islands, and type and abundance of egg predators play important roles in influencing nest-site selection and ultimately the distribution and abundance of Emperor Geese and Cackling Canada Geese.