Intraspecific and interspecific variation in the trophic position of various consumers is central to many theories of aquatic ecosystem functioning and dynamics. In recent years, such issues have been addressed using stable N isotopes (δ¹⁵N) to estimate trophic position of consumers in aquatic ecosystems. However, one needs to correct for variation in baseline δ¹⁵N among sites to estimate the trophic position of any aquatic consumer using δ¹⁵N. We first discuss the steps needed to select a baseline indicator in river ecosystems, where abundant foodweb isotopic data have been published but limited attention has been paid to identification of baseline organisms. We examined within-site differences in δ¹⁵N among primary consumers belonging to different functional and taxonomic groups and found significant differences in δ¹⁵N. Collectors were significantly enriched compared to other primary consumers. We propose scrapers as a baseline δ¹⁵N indicator because they showed low δ¹⁵N values and were more widely distributed than other primary consumers throughout our study sites. Using this baseline δ¹⁵N, we calculated continuous estimates of trophic position of consumers (invertebrates and small fish) for our 87 river food webs. Primary consumers had a significantly lower mean trophic position (2.3) than predatory invertebrates (2.9) and fish (3.5), and these results are concordant with estimates based on traditional studies. However, trophic-position estimates of consumers (both invertebrates and fish) were highly variable across sites with standard deviations spanning up to 0.67 trophic levels, which suggested potential omnivory. Comparison of these trophic-position estimates with estimates based on mean δ¹⁵N of all primary consumers combined (no targeting of scrapers as baseline indicators) suggested that use of a constrained number of groups and corrections for isotopic differences among groups when estimating baseline δ¹⁵N could reduce some biases induced by the use of various functional feeding groups with variable δ¹⁵N values.