Surveys for amphibians along large rivers pose monitoring and sampling problems. We used canoes at night to spotlight and listen for anurans along four stretches of the Rio Grande in Big Bend National Park, Texas, in 1998 and 1999. We explored temporal and spatial variation in amphibian counts and species richness and assessed relationships between amphibian counts and environmental variables, as well as amphibian-habitat associations along the banks of the Rio Grande. We documented seven anuran species, but Rio Grande leopard frogs (Rana berlandieri) accounted for 96% of the visual counts. Chorus surveys along the river detected similar or fewer numbers of species, but orders of magnitude fewer individuals compared to visual surveys. The number of species varied on average by 37% across monthly and nightly surveys. We found similar average coefficients of variation in counts of Rio Grande leopard frogs on monthly and nightly bases (CVs = 42-44%), suggesting that canoe surveys are a fairly precise technique for counts of this species. Numbers of Rio Grande leopard frogs observed were influenced by river gage levels and air and water temperatures, suggesting that surveys should be conducted under certain environmental conditions to maximize counts and maintain consistency. We found significant differences in species richness and bullfrog (Rana catesbeiana) counts among the four river stretches. Four rare anuran species were found along certain stretches but not others, which could represent either sampling error or unmeasured environmental or habitat differences among the river stretches. We found a greater association of Rio Grande leopard frogs with mud banks compared to rock or cliff (canyon) areas and with seepwillow and open areas compared to giant reed and other vegetation types. Canoe surveys appear to be a useful survey technique for anurans along the Rio Grande and may work for other large river systems as well.