Stable-isotope ratios of carbon (C-13/C-12 or delta C-13) have been widely used to determine the energy base of stream food webs, but such use is controversial due to unexplained variability in algal delta C-13. I used published delta C-13 data from temperate headwater streams through medium-sized rivers (0.2-4000 km(2) watershed area) collected during summer baseflows and original data from streams in northern California to analyze energy pathways through river food webs. The analyses showed three important results. First, epilithic algal delta C-13 and watershed area are positively related, suggesting that effects of carbon Limitation on algal carbon uptake result in C-13 enrichment of algal delta C-13 in larger, more productive rivers. Second, epilithic algae and terrestrial detritus delta C-13 values are often distinct in small shaded streams but overlap in some larger unshaded streams and rivers. Measurements of delta C-13 values may be most useful in distinguishing algal and terrestrial energy sources in unproductive streams with supersaturated dissolved CO, concentrations, and some productive rivers where CO, concentrations are low relative to photosynthetic rates. Finally, consumer delta C-13 values are more strongly related to algal delta C-13 than terrestrial delta C-13. The relative contribution of terrestrial and algal carbon sources often varied by functional feeding group within and between sites. However, with the exception of shredders and scrapers, which respectively relied on terrestrial and algal carbon sources, patterns of consumer delta C-13 clearly show a transition from terrestrial to algal carbon sources for many lotic food webs in streams with greater than or equal to 10 km(2) watershed area. The observed transition to algal carbon sources is likely related to increasing primary production rates as forest canopy cover declines in larger streams, although decreasing retention or quality of terrestrial carbon may also play a role. Improved analyses of algal delta C-13 and delta N-15 combined with quantitative study of organic matter dynamics and food web structure should allow the relative importance of these factors to be distinguished in future food web studies.