Natural abundances of stable isotopes in lotic food webs yield valuable information about sources of organic matter for consumers and trophic structure. However, interpretation of isotopic information can be challenging in the face of variability in organisms at the base of food webs. Unionid and dreissenid mussels, commonly used as baseline organisms in lakes, are uncommon in many river settings and can have variable diets, thus making them unsuitable as a universal baseline for many river food web studies and often forcing reliance on more common benthic insects for this purpose. Turnover rates of body carbon and nitrogen in insects are relatively rapid (1 to 50 days half-life). These rapid turnover rates in primary consumers can result in considerable temporal variability in δ13C that rivals that of algae (>10‰ range within a site). This suggests that using primary consumers as a surrogate baseline for algae may not circumvent the problem of temporal variability and the resultant mismatch of sources with longer-lived, slow-growing secondary and tertiary consumers. There are several strategies for reducing the influence of these confounding factors when bivalves with a known diet are not present. These include sampling over large spatial scales and correlating δ13C of consumers with the source of interest (e.g. benthic algae), sampling baseline organisms multiple times in the weeks preceding sampling of larger consumers (particularly in response to large changes in discharge) and using algal-detrital separation methods and multiple tracers as much as possible. Incorporating some of these recommendations and further exploring variability at the base of the food web will potentially provide greater insights into consumer–resource coupling in running waters and more robust conclusions about food web structure and energy flow in these dynamic systems. Copyright © 2012 John Wiley & Sons, Ltd.