Whole-lake additions of dissolved inorganic C-13 were used to measure allochthony (the terrestrial contribution of organic carbon to aquatic consumers) in two unproductive lakes (Paul and Peter Lakes in 2001), a nutrient-enriched lake (Peter Lake in 2002), and a dystrophic lake (Tuesday Lake in 2002). Three kinds of dynamic models were used to estimate allochthony: a process-rich, dual-isotope flow model based on mass balances of two carbon isotopes in 12 carbon pools; simple univariate time-series models driven by observed time courses of delta(13)CO(2); and multivariate autoregression models that combined information from time series of delta(13)C in several interacting carbon pools. All three models gave similar estimates of allochthony. In the three experiments without nutrient enrichment, flows of terrestrial carbon to dissolved and particulate organic carbon, zooplankton, Chaoborus, and fishes were substantial. For example, terrestrial sources accounted for more than half the carbon flow to juvenile and adult largemouth bass, pumpkinseed sunfish, golden shiners, brook sticklebacks, and fathead minnows in the unenriched experiments. Allochthony was highest in the dystrophic lake and lowest in the nutrient-enriched lake. Nutrient enrichment of Peter Lake decreased allochthony of zooplankton from 0.34-0.48 to 0-0.12, and of fishes from 0.51-0.80 to 0.25-0.55. These experiments show that lake ecosystem carbon cycles, including carbon flows to consumers, are heavily subsidized by organic carbon from the surrounding landscape.