Many chemical, physiological, and trophic factors are known to affect ioaccumulation of polychlorinated biphenyls (PCBs) in biota. Understanding the primary factors affecting fish contamination is critical for predicting and assessing risks to upper-trophic level consumers, including humans. Here we identify PCB contamination pathways that could explain within- and between-species variability in fish concentration levels. Three freshwater river fish species (barbel, chub and bream) were sampled at three sites along the Rhone River (France) where fish consumption is partially prohibited because of PCB levels exceeding the European health-based benchmark. The trophic position was assessed using an innovative approach based on stable isotope analyses and Bayesian inference, which takes into account both isotope data variability and parameter uncertainty. The effect of foraging habitat on fish contamination was addressed using stable isotope mixing models. The fish trophic position and PCB concentrations were found to be unrelated while the exploitation of sediment detrital carbon as a food source appeared to be a critical factor affecting fish contamination. Fish length, PCB concentration of the sediment, and individual fish foraging habitat (exploitation of detrital versus planktonic carbon sources) explained 80% of within- and between-species variability observed in PCB concentrations. These results, obtained for species that have overlapping TPs and exploit different carbon sources, reveal that the important factor in fish PCB contamination is not only what fish consume, but also and essentially the feeding location.