Small mammals in boreal forest ecosystems fluctuate dramatically in abundance and 1 possible mechanism to explain these changes is the bottom-up hypothesis of variation in food supplies. Here we ask if variation in berry crops produced by 6 major species of dwarf shrubs and herbs, epigeous mushroom crops, and white spruce seeds allow us to predict changes in the abundance of the red-backed vole (Myodes [= Clethrionomys] rutilus), the deer mouse (Peromyscus maniculatus), and field voles (Microtus oeconomus and M. pennsylvanicus combined) over 13 years (1997-2009) in the Kluane Lake region of the southwestern Yukon, Canada. M. rutilus is the dominant rodent in these forests, comprising 64 of the catch. Overwinter survival is a key demographic variable in all these rodents, and the winter food supply-principally berries produced the previous summermay be 1 key to overwinter survival. We predicted that berry, mushroom, and tree seed crops in year t would produce changes in rodent density in year t + 1. We could explain statistically 78-98% of the variation in May and August abundance of all 3 rodent species with indices of berry crops and mushrooms in the previous summer. For M. rutilus the critical predictor was berry crops of Empetrum nigrum. For P. maniculatus, the critical species were Arctostaphylos uva-ursi, A. rubra, and mushrooms. Spruce seed crops were not significantly correlated with rodent densities or changes in density. A large fraction of the variation in rodent numbers in this ecosystem is explained by a simple bottom-up model of population limitation.