The interaction of large-scale fire, vegetation, and ungulates is an important management issue in Yellowstone National Park. A spatially explicit individual-based simulation model was developed to explore the effects of fire scale and pattern on the winter foraging dynamics and survival of free-ranging elk (Cervus elaphus) and bison (Bison bison) in northern Yellowstone National Park. The Northern Yellowstone Park (NOYELP) model simulates the search, movement, and foraging activities of individuals or small groups of elk and bison. The 77 020-ha landscape is represented as a gridded irregular polygon with a spatial resolution of 1 ha. Forage intake is a function of an animal's initial body mass, the absolute amount of forage available on a site, and the depth and density of snow. When the energy expenditures of an animal exceed the energy gained during a day, the animal's endogenous reserves are reduced to offset the deficits. Simulations are conducted with a 1-d time step for a duration of 180 d, around 1 November through the end of April. Simulated elk survival for three winters (1987-1988; 1988-1989; 1990-1991) agreed with observed data. A factorial simulation experiment was conducted to explore the effects on ungulate survival of fire size, fire pattern, and winter severity during an initial postfire winter (when no forage is available in burned areas) and a later postfire winter (when forage is enhanced in burned areas). Initial ungulate population sizes were held constant at 18 000 elk and 600 bison. Winter severity played a dominant role in ungulate survival. When winter conditions were extremely mild, even fires that affected 60% of the landscape had no effect on ungulate survival during the initial postfire winter. The effects of fire on ungulate survival become important when winter conditions were average to severe, and effects were apparent in both the initial and later postfire winters. The spatial patterning of fire influenced ungulate survival if fires covered small to moderate proportions of the landscape (e.g., 15% or 30%) and if winter snow conditions were moderate to severe. Ungulate survival was higher with a clumped than with a fragmented fire pattern,suggesting that a single, large fire is not equivalent to a group of smaller disconnected fires. The interaction between fire scale and spatial pattern suggests that knowledge of fire size alone is not always sufficient to predict ungulate survival.