The occurrence at different trophic levels of 17 poly- and perfluoroalkyl substances (PFASs), including perfluoroalkyl acids (PFAAs) and some of their precursors (e.g., perfluoroalkane sulfonamides, 6:2 fluorotelomer sulfonate), were investigated in riverine freshwater food webs in Southeastern France. Two fish species (Barbus barbus and Squalius cephalus) and various invertebrate taxa were collected at five locations to assess the spatial variability of trophic magnification factors (TMFs). Particular attention was devoted to sample and data processing to minimize potential biases associated with the TMF determination. Fish were significantly more contaminated than invertebrates (ΣPFAS = 7-1811 vs. 0.9-213 ng g-1 wet weight (ww)). Those from the Rhône River presented significantly higher levels due to high concentrations of perfluoroundecanoic acid (406 ng g-1 ww) and perfluorotridecanoic acid (566 ng g 1 ww) ascribed to an industrial point source. Perfluorooctane sulfonate (PFOS) was dominant at the other sites (range concentration = 3.6-134 ng g-1 ww). Two linear regression models were compared (i.e., Kendall regression vs. Generalized Linear Mixed-Effect Model, GLMM). Results showed that TMFs calculated using the non-weighted Kendall regression were higher than those obtained using the GLMM approach. GLMM-based TMFs were consistently > 1 for C9-C14 perfluorocarboxylic acids (PFCAs), PFOS and perfluorodecane sulfonate (PFDS), indicating their apparent biomagnification in the investigated food webs. Comparatively, 6:2 FTSA and N-ethylperfluorooctane sulfonamidoacetic acid (N EtFOSAA) were less often detected and were not significantly biomagnified, probably because of metabolization. TMF estimates were generally consistent across sites although some PFASs (in particular C9, C10 and C13 PFCAs) displayed higher variability, due to a unique extreme value that may have resulted from the contribution of unattributed precursor biotransformation.< Réduire