Abstract. Distributions of species along a freshwater habitat gradient, ranging from ephemeral pools with few predators to permanent lakes with fish, have been used to infer how predation establishes trade-offs that promote ecological specialization. Larval anurans are said to support the trade-off model, but there are few comparable and quantitative habitat data available to assess this claim. I performed a survey of field biologists to evaluate the habitats of similar sets of species in the northern parts of Europe (40 respondents and 12 species) and eastern North America (30 respondents and 8 species), using a standard set of criteria. For six European species I also had quantitative field sampling data, and found close agreement between survey results and predator densities experienced by tadpoles in the field. Distributions of most species were restricted to only part of the habitat gradient, as expected under the trade-off model. The survey confirmed reports that North Ameri-can Rana species replace one another along the gradient, but this was not true in Europe. European Rana were no different from the North American species in their seasonal and geographical overlap, so the absence of habitat partitioning in European Rana may result from interactions with other species or the special impact of glaciation in Europe. Habitats were unrelated to evolutionary relationships among species, suggesting that changes in habitat evolve quickly. The survey approach was useful for comparing distributions of species, and for generating hypotheses about evolutionary responses to habitat gradients.