Fish are frequently considered the top predator in freshwater food web models despite evidence that predatory birds can impact fish populations. In this study, we quantified bird predation rates on experimental populations of rainbow trout (Oncorhynchus mykiss (Walbaum, 1792)) created by stocking nine small lakes in British Columbia, Canada. Combining estimates of fish mortality with estimated bird predation rates allowed us to partition fish mortality into that due to birds versus cannibalism. Our results indicated that bird predators had significant impacts on age-1 trout populations, but little impact on age-0 trout. Common loons (Gavia immer Brunnich, 1764) were the principle predator among eight predatory bird species present, apparently consuming nearly 50% of all stocked age-1 trout and explaining almost 50% of variation in mortality rates. Age-1 trout mortality did not differ significantly from zero in lakes without loons. Birds consumed a small proportion of age-0 trout, and estimated consumption explained none of the variation in age-0 trout mortality among lakes. We conclude that birds affect fish populations by asymmetric predation on different age (size) classes and can be important top predators that should not be ignored when characterizing freshwater food webs in lakes.